Why AI Agents Are Tools, Not Junior Engineers: Vasu Raj Jain on Staying Technical as a Leader
What we talked about
Vasu Raj Jain is a Senior Software Engineer and Engineering Lead at Amazon Ads, where he owns the architecture for ad-serving infrastructure that handles over 50 million requests per second across Prime Video, live sports, shopping, and audio. He directly leads 30 engineers and influences more than 150 across the organization, and is actively building agentic AI systems. His path into engineering was almost accidental: a childhood dream of becoming a pilot ended over an eyesight test, a month of chemical engineering followed before he switched to computer science, co-founded a hyper-local street-food app, and moved from India to the University of Florida, where a theoretical mindset gave way to a builder’s one.
Show notes
Vasu Raj Jain leads engineering at Amazon Ads without ever leaving the codebase, and he turns that habit into a full playbook for staying technical, building trust, and putting AI agents in their place. He walks through the balancing act at the core of ad tech, where advertiser reach, publisher trust, and customer privacy all have to stay in harmony, and makes a hard case that agentic AI is a tool, not a replacement for junior engineers.
What we covered
- The architecture behind ad-serving infrastructure at Amazon Ads that handles over 50 million requests per second across Prime Video, live sports, shopping, and audio, and why advertiser reach, publisher trust, and customer privacy all have to stay in harmony or the deal falls apart.
- His almost-accidental path into engineering: a failed pilot dream over an eyesight test, a month of chemical engineering, a switch to computer science, a co-founded hyper-local street-food app, and a move from India to the University of Florida.
- How he leads without losing the code, splitting a two-week sprint across writing code himself, designing systems, and mentoring, and deliberately asking his own team to assign him sprint tasks so he understands their execution layer.
- Why he and Federico agree that agentic AI is a tool, not a replacement for junior engineers, since agents run on your identity and your credentials and the accountability always lands on the human who spun them up.
- Why junior engineers bring context, judgment, and a human touch a tool cannot, and why long-tenure teams who hold the business context are what steer agents well.
- His own rule of never letting an agent commit its own code.
About Vasu
Vasu Raj Jain is a Senior Software Engineer and Engineering Lead at Amazon Ads, where he leads architecture for ad-serving infrastructure across Prime Video, live sports, and digital out-of-home platforms, directly leading 30 engineers and influencing more than 150 across the organization. He studied computer science after a brief start in chemical engineering, co-founded a hyper-local street-food app, and moved from India to the University of Florida on a student exchange.
- LinkedIn: https://www.linkedin.com/in/vasujain00
Episode 163 of the PreVetted Podcast.
Full transcript
Federico Ramallo (00:00) Welcome back to the Pre-Vetted podcast, where we spotlight extraordinary people and remarkable talent reshaping our world. Today I am joined by Basura Raj Jain, his senior software engineer and engineer lead at Amazon Ads, where he leads architecture for ad-serving infrastructure, handling over 50 million requests per second across prime video, live sports, and digital art home platforms.
He directly leads thirty engineers, influences a across one hundred and fifty people in the organization and is actively building agentic AI systems on the cutting edge of how ads get served at scale. S Vasou, welcome to the show.
Vasu Raj Jain (00:47) thank you for the nice introduction. yeah, it was really nice to be here and I’m excited about this podcast.
Federico Ramallo (00:54) Yes, me too. Me too. I I I the the way that I see the introduction is so is it it’s my gift to you because it’s a way to, you know at least it happens to me, you know, when I’m building stuff, when I’m doing stuff. I I don’t kind of recollect everything that I’ve been doing. So it’s very nice when, you know, somebody somebody else kind of goes through the the achievements that I’ve been doing that, you know.
ha I haven’t I haven’t realized until somebody else mentioned it because I’m kind of focused on the next thing, right? So that’s kind of my intention with the with the intro, right?
Vasu Raj Jain (01:33) Yeah, you’re right. sometimes it happens that when you are doing stuff, although it might be like very extraordinary or something that is fascinating to somebody else, but for you it’s just like a daily work because you’ve been doing that all your life and you think that it’s just a task. But when somebody else portrays it or tells a story about what you have done, it sounds remarkably like very fascinating and a big achievement.
Federico Ramallo (01:58) Right. Right. So yeah. so anyway, go going a little bit more on the interviews, you know, I I kind of tell a little bit about you know, who you are, but can you tell us a little bit more about what you’re doing today?
Vasu Raj Jain (02:16) yeah, absolutely. So my name is Vasio. I’ve been with Amazon for like more than six years. when I meet people, people do generally ask me a question like, what do you generally do at Amazon? So I ha I sort of like give them a funny answer that I am the guy who is like showing you the digital billboards on everything that you are doing with Amazon, be it prime video, if you’re watching a season series episodes, I’m the one responsible for showing you ads. be it like a shopping experience, if you wanna see like the sponsor products or anything.
I’m the one who is like putting an ads there. Or if it’s like Audible or Spotify or like Amazon music that you are like listening to, if there is an ad break coming, that’s because of me. So people just get like very curious like, how do you make sure that everything is like working throughout the Amazon Amazon like ecosystem? And generally my answer there would be that I’m the lead software engineer who’s responsible for building all the ad tech systems that is integrating with all the platforms responsible for stitching ads. So all the
Like Amazon’s customer journey that happens today gets an ad through the system that my team currently owns, and I’m the lead architecture architecture there. I’m the one who has built those systems. So in a nutshell, I’m I’m the one responsible for like delivering ads through my system to different pieces of Amazon where users are coming and like interacting.
Federico Ramallo (03:40) Right, right. And I I I think that in in the world where we are today where you know we’re oversaturated by information, being able to have ads that are relevant to us becomes much more important, right?
Vasu Raj Jain (03:55) Yeah.
not just important, it’s also very I would say a gray area today. we just don’t want to show a relevant ad. We also want to adhere to privacy as well. like we can’t like really use customers information to target ads. there are like so many laws by different countries, different regions, continents that we need to abide by. And hence there is like a privacy agreement as well that we need to abide by when we are like showing ads which a customer may find useful.
And abiding by those laws while trusting those boundaries and maintaining those boundaries. So it kinda gets tricky. but yes, we try our best and not to like slip onto those gray areas to use like some of the customer information because customer obsession is like our true main like Amazon’s principle that we abide by. So there has to be like that trust from customers perspective to a company. so
Definitely that’s something that we need to abide by.
Federico Ramallo (04:57) Right. Right. I it’s it’s interesting because it’s it’s very easy to explain in a very, you know, shallow level, but then on in all these details with what you’re describing is basically a multidimensional problem, it’s where it gets tricky very fast, right?
Vasu Raj Jain (05:18) Yeah, multi-dimensional as well as like from advertisers’ perspective and the customers’ perspective, right? Me as an advertiser would just like be very interested in showing ads to everybody. Like the broader the reach, the more the chances of selling my product. But when it comes to like inventory owners, like me owning a billboard or a building where I can deliver an ad, I don’t want to show any ads that will let people do not come to my site. I don’t want to show any ads that are like not relevant.
Enough or not good enough for the people who are visiting my site. Let’s say you own a particular site for selling the kids’ toys. Now you don’t want to show any adult ads on that kid’s channel or the site, right? So there are like different dimensions when it comes to like intent as well, like from what a customer’s intent is and the site owner intent is versus what our advertiser intent is. So there are like different moving pieces that we need to abide.
By and we call it rules and like ad safety policies that we need to see and evaluate at the runtime to make sure that we are not like breaking the trust of anybody, like both the parties should be happy.
Federico Ramallo (06:32) Right, right. Because if you go too conservative then the ROI for the advertisers is going down, right? If you go too aggressive you are invading the the user’s privacy. So it’s a it’s a very delicate balance between compliance, and privacy and ROI for the adv advertisers, right?
Vasu Raj Jain (06:56) And all of these things like led to building the trust. And we need to build the trust with all the parties. We can’t like make anybody unhappy because it our ecosystem works when everybody works in harmony. everybody needs to like be at a balance. it’s more like a buyer and seller. If they both are not in agreement, then there is no deal that’s gonna happen.
Federico Ramallo (07:18) Right. Right. Interesting. Interesting. so kind of changing a little bit topic, you mentioned that you wanted to join the military in India, and failed the exam, right? can you tell us a little bit more about that story and how that affected your life?
Vasu Raj Jain (07:39) Yeah, absolutely. So I was always fascinated by being a pilot. throughout my like childhood, I was always dreaming of becoming a pilot. And during my high school, I appeared for an exam called National Defense Academy, where I was supposed to like clear the pilot exam. Although my written exam got cleared, but in a physical exam I failed because of my eyesight. And that’s when it shattered all my dreams of becoming a pilot. And then I j I was just lost. Like what should I do next? because I was preparing for this pilot exam throughout my life. It was
Just my eyesight, which was impacted, and due to that, I’m not gonna be fulfilling my dreams, and that’s how I ended up not going into that work stream. But then my parents told me that it’s okay, it’s not the end of the world. because if that door is closed, there are like multiple different opportunities that you can explore. Maybe you can go and be an aerospace engineer, maybe you can become a mechanical engineer if you can’t fly the plane, maybe you can build the plane. So I thought, yeah, maybe engineering might be a good.
thing for me to like explore and that’s how I started like into the math science field and I started like learning more about the technology and whatnot and that’s how the transition from like going into like defense academy to becoming an engineer happened.
Federico Ramallo (08:58) That’s amazing. I think that the one of the most important characteristics that we can’t that that that you have and I I I encourage everybody to build is resilience and adaptability because we can have a dream b and then for you know i it could not happen for different reasons, so being able to adapt and find the opportunities and be able to move forward becomes much more important.
Vasu Raj Jain (09:26) yeah, I mean absolutely like that was not the one transition that I saw very early on. There were multiple other things that happened during my engineering journey where I switched from chemical as well to computer science.
Federico Ramallo (09:39) Right. I I I had a similar situation where I went through different economical unstabilities. I had in Argentina, you know, the hyperinflation was something that I lived like f three, four times well two two times hyper and then inflation, you know, more th more than five. so you know, after that you kind of get used to, you know,
it’s infl there there’s inflation. It’s like it it’s nothing compared to what I’ve lived before. So it’s kind of okay, right? And and that that allows you to you know, have a peace of mind when there are difficult situations in front of you, that you can kind of navigate through.
Vasu Raj Jain (10:28) Yeah, that’s very important, I guess, because that just like makes you learn how to adopt in this world because there are like so many moving variables around you that you cannot control. What you can control is your behavior and your actions towards those situations which are not created by you. And if you learn that, then that is just the basic skill that is required for you to be successful everywhere. Because things around you you cannot control. You can just control how you react to those things and what actions that you take next to sort of like move ahead.
Federico Ramallo (10:59) Right. Right. I that this reminds me to another story where I I basically broke down a production site for Coca Cola, right? and you know, i it was miscommunication where they told me this is your staging server, but it was also the production server. It was on Windows and I had to install a a D L L library, right? So I had to restart the the server, right? and
Nobody told me that it was also a production server for another project. So, you know, it was just a miscommunication. But, you know, at that time, I was nervous because, you know, didn’t know that I that I did that by the time, you know, people were freaking out and realize it and and and and I basically w was able to say, Well, I did I did I restart that server, I didn’t know. So, you know, then we were able to focus on on on fixing the problem, right?
Which was just waiting for the restart to happen, right? But you know, but at that time i I I was very nervous. I thought I was going to get fired. but then after that, you know, when when I make another mistake or when something happens that it’s you know high stake situation, you can, you know you you know how to deal with that without, you know, getting stressed out, right?
Vasu Raj Jain (12:22) Yeah, absolutely.
Federico Ramallo (12:27) And I understand you also tried chemical engineer, right? For a for a little bit.
Vasu Raj Jain (12:35) I sort of like so
As soon as like I graduated from my high school, I got an admission onto a chemical engineering program. And since I was not becoming a pilot, I thought it’s anything that I can do now because nothing interests me as much as becoming a pilot was. So I thought, yeah, maybe I should just give it a try and like go with the flow, like wherever the opportunity or the door is opening for me, I should just go and enter. And the moment I started like learning about chemical engineering, it was already like a month past, and I asked a question to myself, like is this something that I’m really
Enjoying, or I’m just like here because my pilot dream is not anymore there, and that’s when I realized that all of these chemistry nuances and the formulas and the chemistry lab is not something that I’m enjoying the most. I’m just merely here because I just want to complete my engineering degree. And then one of my friends asked me, like, what do you really want to do? And I’m and I’m like, I don’t know, I’m just like confused, I don’t see myself being here as well. As I don’t know what will be the path next to me, it’s just that I enjoy being in the college because of you.
So he saw he said, why not just like give it a try to all the different engineering programs and see like which fascinates you or which interests you more? And the moment I moment I entered the computer science engineering course, I started feeling that yes, this is something that I can do
rest of my life or I can act actually move forward with because they’re the problems were real problems that I I’m being asked to solve. the logic, the maths behind it, everything was started like making sense to me. And that’s when I switched from chemical engineering to computer science. And it was not like an easy thing for me to do. I had to convince my parents, I had to convince the teachers that okay, I’m okay to like move ahead with computer science. Even though if you will not be giving me admission to this college, I’m okay to wait for next year, but I wanna like start with the computer science itself because
chemical engineering is not something that I can graduate in a and I don’t want my majors to be in that.
Federico Ramallo (14:27) Right. Right. Wow, that’s fascinating. I I I remember when I was eighteen and I had to design my career, I went through something similar. I ended up starting industrial engineering rather than software engineering. because I didn’t know what I wanted and you know, I was kind of drawn into that and didn’t realize how how different it was. But eventually I was able to get into the software engineering you know,
career and you know, it it’s but at at that time I thought I was, you know, in the right direction, right? So it i it it is it is hard to to have the level of awareness and and figure it out. So I congratulations for making that difficult decision at that time.
Vasu Raj Jain (15:15) Yeah, I mean now that I think of I do see that a lot of people are all already in that place as well where they have to make like one choice and that choice can change their life because it’s the career and at that age you’re not even sure about what we you will be doing like next. So it’s kinda like tough thing to go through, but at least like once you decide what you wanna do next, it becomes all easy you for you to follow your ambitions or the passion or maybe the things that you love doing. And that’s what I think I recommend all the people going through the journey. go
your guts like whatever you enjoy doing just choose that as your major you don’t have to like follow others dream or others’ opinions onto your career like if you don’t enjoy something do not pursue that
Federico Ramallo (15:58) Right. Right. Yeah, I I completely agree. It’s I mean in retrospect it’s easy to to you know to see the good decisions, the bad decisions, but when you are there with the limited information and you know, in my case I had a lot of self doubt, and I I I believe a lot of people also have that self doubt. So it it’s hard to make those decisions, right? So yeah.
So tell tell me a little bit more about how you went from, you know, starting in India to then moving to the States, to the United States.
Vasu Raj Jain (16:36) yeah, that’s also an interesting story. So back in my like senior year, there were like different companies coming to my college for like offering you jobs. And at that time there was like a student exchange program going on with University of Florida and I was not even sure about whether I wanna move to states or not. I thought it’s just an exam. I should just appear for it and even though if I’m gonna fail, it it’s not the end of the world because I don’t wanna do that. And there were like five hundred people who appeared for that exam and out of five hundred people, twenty were selected.
And I just got lucky, I don’t know how I got selected for that. And then I thought, should I just like work in India or just maybe like go for one semester abroad, see like how the culture in the US is and maybe come back and then do my job at India. So that’s how when I cleared the exam, I thought maybe God wants me to like go outside and explore how the like studying culture outside of India is. And that’s when I took that opportunity and I thought I gave myself a reason that it’s just a six-month thing. I can just go learn and come back.
Federico Ramallo (17:36) Right. Right. It’s I I I
Vasu Raj Jain (17:39) And then when I moved here,
I started studying my like last semester of my college and I got good grades here. Professors offered me, why not you stay for like your grad school as well? We’re gonna offer you scholarships, we’re gonna offer you a research position in the lab. And that just like started like creating more thoughts in my mind, what I enjoy here versus what I was doing in India. And that gave me like a very different perspective, in a sense that is it really like something
That I should just do here because I I’m enjoying a lot of practicality in the mindset of people as well, in building stuff, in learning more, and exploring the depth of some of the aspects of like human-computer interaction or blockchains or some of the like upcoming technologies back at that time. So that just made me feel that yes, this is the right path for me to explore more rather than working at a company back in India, because that I can do anytime I want. My learning should not stop.
Federico Ramallo (18:40) Right. Right. And once you get immersed into into the country, into the culture, into the university, and you get invited to different projects, they they pull you into to interesting things that otherwise you have you you wouldn’t have experienced or have learned about it, right?
Vasu Raj Jain (19:01) Exactly. Yeah.
Federico Ramallo (19:04) And how was your experience you know it’s is when you go to the States, it’s a whole different culture. What what what has been the most interesting things you find with with this new culture?
Vasu Raj Jain (19:20) Yeah. So
It it’s it’s a very great question. So at that time I was just merely like like guy who’s like just moving from India to US with a mindset that yes, it’s just gonna be a a short trip and I’m gonna be go b going back after six months. The moment I came my first class when I went to the class, I saw a complete change in the mindset of like the professor as well as the students. Because in India the s the teaching culture was more theoretical oriented in a sense that what they are teaching you, be it like algorithms, formulas or
Anything. The expectation in the exam was that you know that algorithm, in sort of like somebody who’s memorizing that algorithm can succeed in that exam. But when I moved to the states, it was not about how well you remembered that algorithm, it was more about what you learned out of that algorithm and where you can apply it at a real world problem. That was like a shift for me from that mindset to diff.
practical mindset and that just fascinated me because to me what people have solved already in the textbooks is not relevant to me when I’m seeing some of the new problems in this challenging world. I wanna see like how what I have learned from the textbooks I can apply around me to solve a problem was the mindset my mindset change that happened.
And that was something that made me think that this is this is what I was missing back in India in terms of like learning experience. And that just changed everything, how I was pursuing my studies, how I was looking into computer science and whatnot. Then I started like seeing around myself and seeing like what I learned in the last semester. Can I apply something to solve some of the problems? And I’ll give you like one, two examples about the problems that I solved back in my university. So at dorms, at like student housing, there is a lot of food wastage that happens.
And the reason why it happens is because people do not often take care of the expiration date of the stuff that they’re buying from like grocery stores. So at human computer interaction, I gave this idea to my professor that what if like I build build an app where when you’re buying something, you just like scan through camera and it will gonna prompt you that milk is in your refrigerator, it’s gonna like expire in seven days. Go finish it or throw it out if you’re not like if it’s already empty. So that whole like even though it was like a very simple solution to a problem, but
But
that just started like solving problems of food wastage. Similarly, there was another problem that I solved was around like people putting like garbage or stuff on a beach. So I created a VR game for people to learn about the turtle’s hatching journey, that when a small turtle gets hatched, how difficult it is for it him to move from sand to an ocean because humans are throwing so much waste on it. And when I created a game out of it, and people being turtle in that VR game, they found it very difficult to cross the ocean, and that the and I said that that’s what you are doing to the turtle.
As well. Like whatever you’re facing in the virtual reality, that’s what a turtle experience is. And that just changed like people’s mindset. And it was like launched in the museum of Florida as well. So those are like some of the real-world problems that I solved by learning some some of the things in the university, and I feel like those kind of experiences for me was lacking in India.
Federico Ramallo (22:24) Right. Right. The yeah, I I’ve seen you know, people trying to change the culture by scolding and I don’t th I don’t think that works, but when you you invite people to experience to to to to experience the problem in this case the turtles, right, then as if if it happens to them, then that’s how you build awareness. It’s a s it’s a more soft and subtle way to do it and and it’s much more effective, yeah.
Yeah, I I I had a similar experience w in my in my university where they basically gave us incomplete problems and we basically had to go and fear out figure things out through the books and the materials and sometimes even it was outside of the materials. We had some few professors that would ask us questions outside of the, you know, official material books. We had to go and and search it out and basically be real resourceful, right? And
and and that that makes a big difference on how you approach problems because I mean there there’s some some things that we we do need to learn by heart, right? But but what’s the point of learning by heart if you cannot apply it?
Vasu Raj Jain (23:43) Exactly.
Yeah, that that’s a great point. And that shift in mindset is like very crucial because the world is evolving a lot and there are like so many problems that are spinning up recently. If people’s mindset would not get changed to see like what they have learnt and apply it at a real world problem, then there’s no point of learning because if you’re learning just to remember something or memorizing something, it’s not gonna be useful for you to apply it to solve any of the problems around you. And then I’m not sure like what’s the point of learning something when you
You can’t even like use that to solve a problem. It’s like you have learnt a language but you don’t know where to speak about about it.
Federico Ramallo (24:21) Right, right. You have all this useful information, right? It’s like knowing you know, the answers of of all the trivia questions, but then you cannot you know, follow up on a project, right? So then what’s what’s the point of all this knowledge if you cannot apply it, right?
Vasu Raj Jain (24:39) Exactly. It’s it’s more like you have all the tools around you and you’re given a task to fix the window, but you don’t know where the window is.
Federico Ramallo (24:48) Right, right, yeah. I I I I know a joke that is related to that is, you know, if if your car breaks down in the middle of the road, you know, the mechanical engineer could design a a new carburetor, a new injection system and can can give you, you know, all the specifications of how the, you know, internal combustion engine works, but won’t be able to fix it, right? What you need is just this you know
mechanic that has practical experience, maybe doesn’t know anything about, you know, the engineering side, but can, you know, fix it with a with a little rope or just, you know, a quick fix and get you going again, right?
Vasu Raj Jain (25:31) Yeah, I I think that’s also a great example. learning by doing things is always like a great experience than just like learning by reading something or watching.
Federico Ramallo (25:41) Right, right. Yes, I completely agree. The the w w one one other experience I I had myself wa when I start working with American companies, you know, the the the culture, so what I what I noticed is that they value the ideas more than the hierarchy, right? So you know, even though I was a trainee when I joined Microsoft, I was sixteen years old.
That’s how how kind of my my s my career got started. I they appreciate the good ideas. And they will tell me if I had a bad idea as well, right? They will give me feedback, right? But even though it was a a a trainee on the lowest level hierarchy, they appreciated and encouraged for me to provide, you know, ideas. and if they were were good, they would, you know, act on it, right?
So I I think that’s that’s the other that that’s another thing that for me h has been a big differentiation on on working with American companies, right?
Vasu Raj Jain (26:48) Yeah, absolutely.
Federico Ramallo (26:51) So I wanted to ask you a little bit more about kind of changing topics, but a little a little bit about your experience going from, you know, writing code, being an individual contributor, to leading a team. How how was your experience transsi going through that transition?
Vasu Raj Jain (27:11) So when it comes to like writing the code by yourself, it sort of like relates your success with whatever you are doing by your hands, right? If I’m being asked to, let’s say, cook something and I’m a good chef, I can cook it and I’ll get like appreciation that yes, Vasu has done something very great. But with when it comes to being a lead chef and you have like six chefs around you and you need to like train them, make them also cook a good meal for your guests, then it’s a job of like their success is all
also tied to your success and same thing happens with like when you are writing code by yourself versus when you are like leading a team and engineers and mentoring them as well as to make them successful their success is tied to your success so it’s kind of like a like a
Like a trust boundary where you need to like sort of like learn and do things by yourself, but also enable path for others around you to deliver so that they can also get succeed in their role. And then their success is like like relatively proportional to your success. So it’s more like a teamwork where you need to help and block others and get things done through them and help them achieve success and mentor them where they are they are not like good at, or maybe there’s an area there they can improve or they need like some help.
with. So it’s it’s it’s kinda balance that you need to maintain. you can’t like just do everything by yourself when it comes to a big
Companies or the teams, you need to empower people around you so that they can learn things and they can also work at the same pace that you were performing back then. So, what I generally do is like I follow this tendency of like dividing my whole like two-week work plan with 20-30 percent of the code writing by myself so that I also know the groundwork, what is happening in the code bases and around what people are writing. 30 to 40 percent I spend my time in designing the systems, right? And remaining time I act.
Mentor the people around me so that I I know what the gap areas are because I’m also adding code to the groundwork repositories and the packages where I know these people have committed code so that I know where the area for them to improve is. If I see that somebody is writing a code but missing a unit test, then I coach them there that okay, this is where you need to look at when you’re submitting some of the code changes to do. Look at all these corners, look at all these cases that can happen. I give them ideas to look at what if we have to do it this way, so that they can also.
Start thinking in the right mindset that I am thinking because if you have to like make something successful, you need to make people around you also be thinking in the right direction and also empower them to think out of the box. It’s it cannot be that you are the one who are doing everything and these people are just merely following you. You want them also to think like a leader, you want them also to take responsibility, grow people around them so that all of this thing is like working in a harmony to deliver something very big.
Federico Ramallo (30:06) Right. Right. It it it is it is a delicate balance and it’s it’s it’s I I I believe that a lot of new managers struggle to figure it out. so I I I appreciate how much clarity you’re you’re providing and how much percentage you’re spending your time on. the I I think that mentoring people and helping your team to do better, it’s it’s so important.
Because if you try to do it all yourself you become the bottleneck without realizing it, right? And if if if you don’t lean on your team then you know and you don’t build that trust then it’s hard to to be able to be effective as a team, right?
Vasu Raj Jain (30:55) Exactly. And it it’s it’s a two-way communication. like generally you you have to be like open enough to understand the other’s perspective as well and be appreciative about it as well as like look into where the areas are for them to improve and help them out in building that trust by fulfilling those areas of improvement in a way that they don’t feel obliged about that they are not good at it, but they are also certainly learning and thinking that yes, that’s something I would have to like
work on a or improve and do it in a way that they don’t feel offended about it as well.
Federico Ramallo (31:31) Right. Right. one of the things that I found out when I moved to Mexico is that the the team required the team that I was managing required more of my effort to get the feedback t for th for them to provide me feedback. They were afraid of providing me feedback directly, so I had to build systems around it so they could provide anonymized feedback
We could have open discussions about about things because I mean in retros I I I I mean it’s the same thing as we talk with we were talking about before, right? We make decisions and in retrospect then you realize, it was a good decision, it was a bad decision, right? So I I didn’t realize at that time that I would put people on the spot on a public meeting asking for feedback, right? And people would would get nervous and I, you know, didn’t realize that I was making those mistakes, right?
Vasu Raj Jain (32:30) Yeah, I mean that’s that’s a good example, yes.
Federico Ramallo (32:35) Yeah, and and then the other thing I I learned through through my experience is that it that it’s you know, managing a team it’s a it’s a whole philosophical different problem than managing code, right? When you’re writing code it’s you and the computer, right? It’s a it’s a think problem and the other one it’s a people problem, right?
Vasu Raj Jain (32:57) Yeah. Yeah.
That’s where I think what have worked for me is that when you sort of like understand and be in the shoes of what the other person is currently doing and understand the ground up work as well, then it becomes much more easier to understand the context that they are bringing while talking to you, as well as you also understand what is happening at the bottom layer. Like generally things where it it it is it is not working out for people is because they are always looking out from the very top level of the problems. They’re not understanding or putting
themselves onto the shoes of the junior engineers and looking into what is happening at the at the reality level at the ground level. So that’s why what I do is like I spend like 30 to 40 percent of my like time doing the work that they are also doing so that they don’t feel that I don’t know what is happening in the code layer. So I spend my time taking the sprint task, I ask them to assign me some work so that I can understand their mindset, like how they’re breaking a task, how they’re thinking about solving a problem. And that’s where I get those data points that I need to coach them here so that they can
make better decisions and it’s it’s it’s a it’s a continuous journey like you will not be like seeing results positively right away but at least that helps in building the trust that okay somebody who’s leading me is also well aware about how to make a change somewhere which is like very very much like a small thing but at a junior level but he’s also appreciative enough to pick up those tasks it’s not that I’m always gonna like design systems I want to give that opportunity to the people around me as well so that they can also grow in the ladder and become like senior
Federico Ramallo (34:22) Right.
Vasu Raj Jain (34:29) in the team. So what I do is like I spend my like 30 to 40 percent of good amount of time in taking the task from these people so that I can understand how their execution layer like mentality is like are they able to execute project efficiently are they thinking in the right direction are they giving us a good task are they looking into the corners of or not so that just makes me build trust with my team teammate as well that okay this guy he doesn’t own always like show up in a meeting or design systems he always he also writes
He also writes the same amount of code that we are writing. So it’s it sort of like builds a trust with Julia engineers, and they’re like very much happy about it.
Federico Ramallo (35:07) Right, right. You you’re building trust by leading by example.
Vasu Raj Jain (35:17) Yeah. Or or just being in just doing what they are also doing so that they feel that yes, this is how a leader should operate. He should be ready to like
write code anytime that we want him to and he should be appreciative enough to take tasks from the sprint plannings or or like that. So I always like ask that if you guys have any task for me, give me some tasks from this package and that package and I’ll be able to do that. And that’s where like I find some of the gap areas and I coach them that why we are not doing it this way. And then they learn about my mindset and they think that okay this is what we should also be looking into. How to improve some of the operational burden in our team by looking into the corners and so that we don’t repeat
make the same mistakes again if we are building something new.
Federico Ramallo (36:00) Right, right. Interesting, interesting. Ha h how do you deal with the because what happened to me is that the amount of interruptions increased and my productivity as an individual contributor went down. because you know, an interruption I would have to focus again. how how do you manage that?
Vasu Raj Jain (36:22) so right now I would say that interruption has gone very minute because now that I’m so surrounded with n number of teammates and n number of agents around me. So I have like so many agents that are using my persona to do the job, and that’s how I’m able to like get my coding things done through agents around me as well as be responsible for attending the meetings where in-person attendance is important and my feedback is important. So Agentic AI has helped a lot in removing that obstacle of interruptions because
Now that I’m when I know about a task, I can create a spec anytime I want. I don’t have to be doing it during my day hours when I’m attending a meeting or doing something where I’m responsible for. whereas these coding tasks, I can spin up a job where my agent is writing a code using my persona understanding problem, prompting me that hey, this is what we should be doing, and I’m reasoning with him. Just like how a pair programming works, that that has just like solved the problem of interruption for me. lately, when the agentic AI or these coding agents were not there, at that time.
it was pretty challenging for me as well to be like figuring out a path forward to deal with the interruptions but what i used to do what i i used to have like my own like focused hours dedicated in a day which were around afternoon because in afternoon there were very less meetings so that’s when i used to like write code but i used to like divide my calendar on a daily basis of like three hours of my dedicated time where i’m not gonna be part of any meetings and i’m just being responsible for delivering the tasks that are that are assigned to me where
versus like managing other times to help out others and mentor other people.
Federico Ramallo (37:59) Right, right. That’s amazing. I I I I’ve been experiencing more and more the agentic development and I was I was thinking on on on talking about that, but I was kind of focusing more on the people problem first because you know adding agent agents built builds another layer of complexity to managing a team, right? but anyway, I think I I I agree with you that
you know, because the usually what happened to me is that I I lose context, my personal context, right? Like what was this ticket about? What you know, I had some ideas, you know, and with having an an agent that you can pair program with, then the agent has the context. So you can ask where w where we we’re at, what were the pending questions or pending decisions that we need to make, right? Can give you a
Vasu Raj Jain (38:51) Exactly.
Yeah.
Federico Ramallo (38:58) a very high level, you know, overview, you can click it again and just, you know, guide the agent towards build what you need to do, right? and rather than spending a lot of time on syntax, then you can focus more on the intention of the code, right?
Vasu Raj Jain (39:07) Exactly.
Exactly. Yeah. And that’s a good example of like how agent tic AI is like s like helping out humans in getting t stuff more done. Like earlier our productivity was like hindered by interruptions, as you said. Like we were like spending time in understanding where we interrupted, what was the context at that time when I paused a job or paused a work or stream, right? Now that we have agent AI, these agents can remember where they were like left alone or they can pick a context from a prior understanding.
And start getting things done at a supplemental level again at the later stages. So this is like a very good example of like how agentic AI is not like replacing humans, but all but helping out humans and getting more things done. And that’s what like I tell a lot of college students these days, where they come up with a question that now that AI can write a code, do we really need humans to do this job? And and my answer to Tim is always like this: that agents will not replace humans. They are there to make humans do more work and do things which are like
Should be done by the humans. Like whatever was the basic chores of remembering things, managing the bookkeeping task and whatnot, that all can be done by the agent. An agent can write a code on your behalf, but the intent behind the code should be given by a human. Because the agent doesn’t know what a system is doing at a real time or like what it should be doing at that moment. You should be the one directing that agent to do the right things, and that’s where engineers are needed.
Federico Ramallo (40:39) Right, right. I I completely agree on I I I think I mentioned this to you before. I I’m almost finishing my book about managing cross border engineering teams and one of when I when I talk about AI, I what I say is that context, human context, long tenure teams, i it becomes much more important than the leverage you can get from agents because
The team with long tenure knows everything about the business and can guide the agents and correct the agents towards the right direction much more efficiently than a new team, right?
So w what has been your experience between managing agents and managing junior developers? I I’ve seen a lot of people making those comparisons. what are your thoughts on that?
Vasu Raj Jain (41:43) yeah, that’s a great question. So managing agents is like a
Is an entirely different thing. I don’t compare that with managing junior engineers. Junior engineers or the interns that are joining a team or a company these days, they are just freshly coming out of the college. They have a lot of theoretical knowledge, they know, they know, they have the capability to think out of the box. They can understand a lot of like different different aspects and bring a different perspective to the team where agents are merely I would say a tool that we are using to get stuff done. So they are not something that we should be comparing like Apple students.
Apples, it might be a bad example to do that, but for me, like agent management is like totally something that I can do on my own based on my intent of like how agentic AI I need to use at a particular thing. Whereas when it comes to junior engineer, in general, the interaction always involves with like making them feel welcomed in the tech world because it’s their first job. probably like for 90% of them, it would be their first job, right? So you need to be like very supportive enough to let them like delve into the
technology world understand what is going on how things work how the industry best practices revolves around them and how they can succeed in that role so it’s more like a mentorship kind of thing that I do when it comes to junior engineers joining the team because I don’t want to feel I don’t want them to feel that it’s like too much for them into this whole new world because that will just gonna ruin their experience of like learning and growing so I mostly like start with making them feel welcome I introduce them to the team making them
learn about some of the best practices that they might not have learnt in the university like how the CICD deployment works, how like staging works, how to test a particular change locally, how to do things which are like very basic enough, but that just gives them confidence that okay they are able to pick up pick it up quickly. I don’t want them to
solve like a NP heart problem on a day one. I want them to understand the basics so that they feel com confident and warm that yes, this is something that I can understand, I can do. And that just like makes them feel more confident. And once they feel more confident, they are able to take more like ambiguous tasks in the team.
Federico Ramallo (43:53) Right. Right. you you can start with smaller smaller tasks and then build from there, right? yeah, I w what I do with with the teams that I I I I manage and I coach is I work towards responsibility. So they they have a lower level responsibility, the juniors where they may not take the ownership of a whole feature, they would take just backfixings or you know
a smaller scope of of task and responsibilities and then the pair programming and the co reviews can help us guide them before they they push they push code that that could potentially be harmful for the project, right?
Vasu Raj Jain (44:42) Exactly. Yeah. Yeah, th that that’s a way to go, right? Because these engineers are fairly like new in this world of tech and we don’t want to like surprise them by giving them a problem that they just fail at. So we want to like start slowly, build the trust, build build the enough capabilities on their plate so that they can also succeed and they can also learn. And once they start learning, they’ll start looking into those things and they can start taking more responsibilities.
Federico Ramallo (45:08) Right. Right. the the comparison I’ve seen I’ve seen doing between junior engineers and agents is that if you think an agent as a new hire, that you can the more clear the requirements you can provide and the workflow that they have to do, the more the more the quality of the output is going to be, right? and I try to explain this to a non-technical people
That way they you know they understand they understand the agent as functionally as a as a as a new team member, right? Where we have the similar interaction between an agent and the rest of the team. and we can think delegating tasks to an agent as as as as you know I’m supervising the agent’s work, right?
but the I I think that the biggest differentiation that I’ve noticed that I think is important to make is the we cannot make agents accountable. The accountability all always relies on humans, right? So it’s it’s up to the humans to supervise the agent’s work, right?
Vasu Raj Jain (46:29) Yeah.
Not just account of like accountability, that also like that is also like a very key factor or differentiator between a junior engineer and an agent, but also the trust, like and observability. Like with a human, there is a human emotional context as well when you’re talking to them, understanding the whereabouts of like things are working. But with agent, there is no such emotional boundary or the cognitive intelligence present in that agent. He’ll just merely do whatever you will say. if it’s in auto mode, it will do.
Whatever it will take to get the job done. But how do we know it is being done in a right way or wrong? is something that we can’t really.
Be sure about if a human is not involved in that process. And that’s where I I sort of like think of agent more as a tooling, not as like a replacement of junior engineers. They are merely a tool, just like how you use ID to write code. Now you ask agent to write a code, that would be like a fair comparison versus junior engineer. He brings the expertise, context, and the human touch into an engineering job, which agent won’t
not be able to bring at least at the moment.
Federico Ramallo (47:43) Right, right. Yeah. I I I like your your idea of thinking the age is more of a you know complex, you know, more smarter tool, but at the end of the day it’s still a tool, right? so it’s up to us to guide it to to to what to to our goal.
Vasu Raj Jain (47:55) Exactly. Yeah.
Yeah, yeah. If something breaks, you can’t just say that agent broke that. the pe the person that agent would be the responsible guy for that outage or something. So that
Federico Ramallo (48:06) Right.
Vasu Raj Jain (48:11) That’s where like you get an answer that is agent and junior engineers are safe. No, they are not. Agent is a tool. like if agent broke something, you will go after the person who was working with that agent or spinning up that agent or giving instructions to that agent. You will not like and if the same thing happened to a junior engineer, then you’ll ask him why that did happen. so those are like very two different things.
Federico Ramallo (48:33) Right, right. Yeah. I I I I th that’s why I I I start talking about no accountability for agents because it’s you know, at the end of the day it’s you know, yeah, it it’s it’s up to us humans to to guide it, right? Yeah.
Vasu Raj Jain (48:41) Yeah. Yep.
Yeah. I I’m I’m
still like very surprised or maybe like fascinated by like living in a world where agents will also have an identity and that that time you will just go after the agent that agent X did this and he’s responsible for breaking something or doing something good. But today agents are using humans identity. anything that you’re spinning up by yourself, that agent belongs to you and he’s using your identity, your credentials to get things done.
Federico Ramallo (49:15) Right. Right. Yeah, I I when I’m using agents now I I tell tell the agents, you know what, don’t make any commits. I will make the commits because that way I’m forcing myself to actually go through the code and review it and make sure that it’s kind of okay, right? Yeah. Yeah, or or or reviewing the pull request, but then by the time you get to a pull request it’s so much code that I feel that, you know, I could miss stuff, so
Vasu Raj Jain (49:31) Yeah. That’s that’s the way to go, yes.
Federico Ramallo (49:45) I’d rather do it on a you know, multiple multiple times a day throughout th before committing, right? And yeah. so, you know, we’re we’re almost running out of time, but I want to ask you, you know, a question about advice. What advice do you give to software engineers that are starting to get or planning to get a leadership role? what advice would you give them?
Vasu Raj Jain (50:16) Okay. so software engineers who are planning to move towards more towards like a leadership role, the one advice that I give always to them is that do not lose the touch of writing code. Do not lose the touch of looking into the groundwork of what engineers around you are doing. Once you like start acting like somebody who can take a job or
Do the work that these engineers around you or whom you’re managing are doing, then it just like automatically builds the trust with them because they know that a person who is leading me also is well aware about some of the nuances in the code base and the things that are happening at a reality level at the at the ground level, and that just like changed the shift of like how that leadership role is supposed to be, and then it becomes more like a trust, trust relationship where people who
Whom you’re managing, trusted, trust you as well. and directions or instructions that you are giving or leading a team. it just works bidirectional. That you give them instructions, they’ll follow, they’ll they’ll learn about it and they’ll understand because they know that it’s coming because you have also seen what is happening there. And that’s where I think a lot of people do not follow this and they just like start operating at a higher level.
Where they don’t know what is happening at a ground level and they don’t understand the complexity of the code base or things. And that’s where the trust starts breaking because a junior engineer starts mistrusting your the senior engineer that he doesn’t understand the code base because he has never touched that code base and now he’s telling me to change X thing. Whereas he doesn’t even know whether that X thing real is already existing or not, right? So you don’t want to be in that situation. You want to be somebody whom these junior engineers can trust and they know that this guy also knows that much that I know.
And that’s why he’s like leading this team. So always tr starts with like building that trust by being in their shoes as well as like taking the task that they are doing so that you also know what what is in their mindset.
Federico Ramallo (52:02) Right.
Right. Right. I I think that being able to show by example that everything you’re asking them to do you also has the capacity to do it and the willingness to do it. I I I I think that that allows you to build a lot of trust with the team because they they can see you as another
a another peer, right? And that they’re going to be more committed towards the work that needs to be done, right?
Vasu Raj Jain (52:50) Exactly.
Exactly.
Yeah, leading by example, insisting on high standards, that way the teammates that you’re leading would follow that high standard role as well. If you are yourself making so many mistakes, then how are you gonna set examples for others to follow? So definitely like how you how you expect your junior engineers engineers to follow, you need to show them that that’s how you also operate, so that they can also like idolize you or see that this is what I want to perform so that that’s when I’ll I’ll also be able to like get promoted or be in that role.
Federico Ramallo (53:24) Right.
Amazing. I think I have another minute for another question, but wanted to ask you what advice would you give somebody that is w you know, thinking to move from India to the United States,
What what advice would you give them?
Vasu Raj Jain (53:44) The best advice that I can give to somebody who’s like planning to move from India to US is that embrace yourself for a big change and that the change would be like a good change for you because then your mindset would change from theoretical knowledge to a practical knowledge. But a cultural change that you will expect is something which is like very personal for everybody to like experience. it might work in your direction that you start enjoying it and it might not be, but that’s something that you can control by controlling the surroundings.
around you. But in terms of learning, the motive that you’re coming to United States, it drastically depends on how you sort of like are looking into learning aspect.
If you are more of a builder kind of mindset, then yes, it will work a lot in your favor and you’ll get a lot of amount of chances to build something. Learn by doing things and a practical knowledge about things. But if that’s not something that you have done in the past and you’re not comfortable with, then I would say that’s where you need to work a lot on. Because here, merely l remembering some something or memorizing something will not work out for you. You need to be of a builder mindset that if you are given a problem to solve, you need to think about different aspects out of the box to solve, come up with a solution of it.
It might not be something that you can look into the textbook textbook and figure it out that this is the solution of for this problem.
So practical mindset is something that people have to build if they’re planning to come from like India to US. And that’s like one advice that I would give to people.
Federico Ramallo (55:03) Amazing Basso.
Vasura running out of time. I truly appreciate you being here today. and thank you very much for joining us.
Vasu Raj Jain (55:19) yeah, thank you for your time as well. it was so nice talking to you.
Listen on your favorite app