Subscribe
Share
Share
Embed
Software engineering isn't just strategy and sprint plans. It's ambiguity, pressure, trade-offs, and people.
Thriving In Engineering helps you navigate all of it.
In the newsletter, Alex Ponomarev distills 25 years of engineering leadership into practical frameworks to help you lead with confidence, stay calm under pressure, and build thriving teams.
The podcast goes deeper, talking to engineers, leaders, and builders across the industry about what AI transformation actually looks like in practice, past the headlines and hype. One guest at a time.
So today, when when we see in our team who is the most vulnerable, mid level developers are the most vulnerable. A lot of the times, AI will produce something that's absolutely amazing, and it, you know, saves you, like, 70% of your time. A lot of the time, it it you'll be wasting 30 more. So if something that was hundred hours, you'd end up doing in a hundred and thirty hours because you were wasting thirty hours on actually looking at code that was crappy. I still hope and I pray that the human race is going to come up on top of it.
Speaker 1:People will have to learn and unlearn and relearn because that's that's the survival game right now.
Speaker 2:Alright. Welcome to our first ever podcast episode of Thriving Engineering, the podcast for engineering leaders who want real answers, not textbook theory. I'm your host, Alex Panamorov, and each episode we sit down with leaders who have been in the trenches, building teams, shipping products and navigating the kind of chaos that no playbook fully prepares you for. If you're ever carried the weight of a team without a map, this is the show for you. Joining me today is Shamim Rajani, tech entrepreneur, COO and co founder of Genentech Solutions, and one of the most distinctive voices in engineering leadership you'll hear from Pakistan.
Speaker 2:Over the past two decades, Shamim has built and scaled software teams across two continents, from Karachi to Detroit to Toronto, working across AI, IoT, cybersecurity and progressive web apps and mobile apps. She is also the co founder of Code Girls, a bootcamp that has trained over 2,000 women in Pakistan's tech industry and ex chairperson of Pasha, Pakistan's only IT industry association. On top of all that, she writes a weekly substack newsletter called The Leadership Lens, where she gets honest about what it actually takes to lead, the self trust, the delegation and the hard conversation most people avoid. Shamim, welcome to Thriving in Engineering.
Speaker 1:Thank you, Alex. Thank you for having me.
Speaker 2:Thank you for joining. For our listeners who know very little about you, you can tell a little bit about Genentech, how you built it. I know that you've been doing this for twenty years. Tell us a little bit.
Speaker 1:Sure. So, I signed off as a freelancer back in 2004. I was working off of freelancer.com. It was called Get A Freelancer back then. And then, you know, the kind of work that I was doing, I was able to get get some retailer accounts.
Speaker 1:So then in 2006, I built a small team, and we registered the company by the name of Genetic Solutions. And that's that's where we started off. Today, we have a team of around 100 plus people. We sit the back office sits in Karachi. The front office and the marketing wing is in The US, in Detroit, And we build solutions for small and medium businesses in the North American and the European market.
Speaker 1:What we do what genetic solutions does is that we build software solutions like, you know, if you have an existing web application or mobile application and you need ongoing maintenance or if you have an existing system or if you wanna build something from scratch or if you're looking to improve your productivity and efficiency using AI, that's where we come in. So whether, know, that's as as consultants or as engineering team or, you know, somebody you might need to do an audit of your existing system. If if you call us, we are like we we do core we we we focus on core engineering. Along with that, obviously, we do also have teams that help in on the design aspects, the on the DevOps and the cybersecurity as well as content marketing aspects. So we have a complete software development site life cycle going in.
Speaker 1:So if you bring a project to me, you can very well forget about every everything, anything around that from project management to, you know, consulting to subject matter experts to deployment to post deployment support. All you need to remember, though, is the marketing part. I don't do marketing. So
Speaker 2:It it's a huge structure most likely. Are you the only manager in the team, or you have, like, multiple layers? How does it work? Because some teams have flat structure.
Speaker 1:So we have a flat structure in in in terms of when you're building projects. So we follow agile, and we we follow scrum mostly. So we have product owners. We have a development team. We have a scrum master, and that works really well in in regards to, you know, delivery and, you know, making sure that the customer gets what he actually he or she actually has thought about and not not not get a surprise at the end of it.
Speaker 1:So that works really well. In terms of our own company, obviously, we do have a a hierarchy, not a very, very not too many verticals, but we do have a decent hierarchy. I do have project managers. I do have product owners that work under us. I I do have a director engineering who works with them.
Speaker 1:And, you know, as the company grew from a one person to a 10 person to a 40 person to a 100 people company, we had to sort of, like, build in different tiers. So it it all came out organically. So right now, the company has a you know, has partners. So I'm not the only owner of the company. I have two more partners now.
Speaker 1:And then we have a team we call the the p six. It's called the Pinnacle six. That's that's our advisory. That's our go to die you know, board sort of like where we go to. These are, like, senior most people from the from the from the company, whether they're directors or SMEs or, you know, super senior engineers.
Speaker 1:They sit in that team, and we we take advice from them at least three times a month. And then we have directors. We have a couple of directors. We have a director of engineering and director of learning and development. And then under them, we have different departments, and then there's a team of project managers or product owners.
Speaker 1:And then there's a team of HODs, which is, you know, for each department has an HOD, and then there's so I I mean, once you now that you're asking me, there is a very decent hierarchy there that follows. Obviously, I can't do everything on my own. So I think one of the things that I learned the hard way and organically is delegation.
Speaker 2:That's definitely the bread and butter of running any team really, especially 100 people. I'm gonna slowly segue into what I want to talk about today. And, you know, for for a lot of engineering leaders, a lot of companies, AI is top of mind. And I don't want to be too optimistic about AI. You know?
Speaker 2:We don't need 100 people teams anymore. It's gonna be one person company here and there and so on. I think truth is in the middle. I also don't want to be too skeptical saying that, you know, last year, a lot of people were saying that AI can't can't do this, can't do that, and look at us where we are today. A lot of things are possible now.
Speaker 2:I'm curious how you see the current state of industry, how you're adopting AI in in your company? I'm just curious about your point of view as a leader because there's there's a lot of a lot of aspects to it, and I wanna I want to hear your opinion, and then I can ask you about, you know, some more details.
Speaker 1:Thank you so much. I think that that's a very that that's the that's the most that's the hottest question you get whenever you enter into any any sort of a podcast. And lucky for us, you know, when COVID came, we were sort of, like, partially prepared for it because even though we had the back office in Pakistan, and Pakistan is a is a it's like conventional conservative society where, you know, you think of you there is you don't think of work life balance, and you just talk about, oh, you have to come to the office nine to five. Even in that at that point, we had a work from home culture, a a hybrid work from home culture. So for, you know, senior people, for women, we we had given that opportunity to, you know, to especially those who had children to be able to, you know, work from the office and then go home and finish their work.
Speaker 1:So we had that infrastructure in place. So when COVID hit, I think we were one of the first companies in Pakistan to actually go, you know, work from home in the next week. I remember it was March 2021, and I had just come back from The US. I was in Qatar, and Qatar was isolated. There was there was not a single like, there was there were bit I wouldn't say there was a little there were very, very few people on the airport, and you know Qatar is one of the busiest airports.
Speaker 1:And when as soon as I came to Pakistan and we went, you know, work from home in the next couple of days, even before the government announced a work from home. So, you know, we we were, like, three weeks ahead of everybody else to go from work. So even with the same goes with AI. So we've been doing some form of, you know, AI work for some of our customers in the hospitality industry for a decade. So we've been using Dialogflow and, you know, that that sort of technologies to be able to build solutions and in the hospitality industry.
Speaker 1:So when the AI bubble actually started and it came we we saw AI coming in two years ago. One of the things that we decided to do very early on is to build a research lab. So we built a lab called the genetic emerging technologies lab or to get labs. You might have heard about that on my, you know, newsletter as well. Actually, I actually about it.
Speaker 1:A and I actively write here, so I'm not sure where I posted that. But so and we started hiring. We put a a senior person in charge, and then we started hiring fresh graduates out of university with a lot of AI knowledge because that's what they were learning. That's what they were, you know, craving, and they were very, very, you know, hungry about. So we we started hiring those people, and then we just spent a year of six six to eight months, I think, or around a year to just research and pump in money there and hire people.
Speaker 1:Just in 2025, early twenty twenty five, we gave the team the the the task to ensure that, you know, by the end of twenty twenty five, we're actually doing a good chunk. We didn't give them a number because we didn't know the number. Because Mhmm. AI was ever evolving. If I had said 20%, I'd be wrong because I don't know.
Speaker 1:So we said a big chunk of what you're doing has to be done by AI by the end of twenty twenty five, and every department needs to to start to adapt. So we're not downsizing, but we're not going to grow the team.
Speaker 2:That's what That makes a lot sense.
Speaker 1:You don't if you don't adapt, you have to leave. So and then then we gave that task to GetLab to not only, you know, supervise that, oversee it, but also to help them and to guide them. So if if the GET Lab went to all these departments one by one and asked them, you know, what their bottlenecks were or what took the most time that was run, and then they did did not, you know, sort of, like, enjoy it, but it was it had to be done. And they sort of, like, found tools to be able to AI tools to be able to replace that. And that's how by the end of twenty twenty five, we know now that our team is actively using different tools, AI, both generic and agentic, to be able agentic more now, generic right from the beginning of 2025, to be able to get a lot of tasks done.
Speaker 1:Now the target is just, you know, a couple of hours today, we were actually having the p six call, and we said, okay. Now the target is now that we have experimented with so many different tools and now we know what tools we like to use and that are also, you know, like, companies that are actually building those tools actively, and we are enjoying using those tools. So one of the tools I'll tell you that we're actively using is Copilot. So we initially started, you know, with a free version then, you know, limited licenses, and now we are actually actively so we also tried others. But we now we're planning to stick to Copilot for now, and we're actively buying more licenses.
Speaker 1:Right? So now the task for each HOD is to ensure that we actually get between 20 to 35% of work done with AI and 20 to 35% of reduction in resources, not just, you know, not so whatever revenue we have this year, if we are going to grow, we are going to ensure that we reduce the human resources. One of the things that, you know, is very I mean, it sounds it it sounds bad. But one of the things that we know about humans is that unless and until we put put them under a pressure test, they don't shine. So what we're trying to say is that, you know, okay.
Speaker 1:So if we have a if we have a development team of, say, 50 people, let's cut them down down to 40 and say, okay. These 10 people, we're giving you 10 agents. Use those, but use them because you still need to take out the same amount of work that 50 people were taking out, but not right now, now just with 40 people. But we've trialed and tested for a year to see it would that it works and that you can. But if you you won't unless and until we actually take that leeway or take that leverage away from you and tell you, okay.
Speaker 1:This is what it is. I don't know. So this is this is, like, honest. Right? This is honest conversation.
Speaker 1:This is what we are trying, and this is what we feel that we might not achieve 100%. But from the experiment in 2025, I think that we might just achieve a big chunk of it. Not all of it, maybe, but a big big chunk. So that's how we are actually working around that. And we consider now AI, and you're very right.
Speaker 1:People are very skeptical about it. I was one of the big the biggest skeptics. Like, you know, no. It's not it's not that bad. But now that I see the way AI is growing and AI is taking up everything from, you know, household stuff towards everything, I think that, you know, we we can't be we we can't be closing our eyes to it.
Speaker 1:So today, when when we see in our team who is the most vulnerable, mid level developers are the most vulnerable. And this I've been saying for, you know, four to six months now that, you know, juniors will need because they bring that abundance of fresh AI knowledge. Seniors are the ones who who actually understand it, and they'll know exactly how to place it and how to guide the juniors. So and most of the grant work, yeah, I can do. So so that's the middle layer that, you know, we see that that that is depleting, that is going thinner and thinner.
Speaker 2:That makes so much sense, and I can relate to a lot of things you said. And you also answered a lot of questions.
Speaker 1:So you
Speaker 2:I I I only wanted to ask you. No. Thank you. I think a lot of engineering leaders today are asking themselves, for example, where do we get this AI talent? Right?
Speaker 2:And I actually agree with you that you can't hire that AI talent from the street, from a job posting. You can't get university grads, the the best ones who are keen to learn, keen to experiment. That's how you solve it. But you can't rely solely on that. You also need to train your your existing teams, your senior engineers, and that takes a lot of time.
Speaker 2:As you said, you've been experimenting for 2025. We have a similar process going on, and I 100% agree with you that mid level engineers are endangered species, and they have to adapt quickly. I've been saying this for quite a while now as well. I think a lot of people on Twitter and on Substack are writing about this as well. And I'm actually surprised by how many mid level engineers are ignorant to the problem.
Speaker 2:I'm not saying everyone. It's just the numbers are staggering. Because for exactly the same reason, juniors, they are extremely eager to try things out. Copilot, Codex, Cloud Code, all of these tools give them superpowers. They're like, oh, now I can build this.
Speaker 2:I can do this. I can do that. It doesn't work half of the time, but they're still happy. With seniors, it's, you know, same thing, but the opposite. As you said, they know what they're doing, and they're like, oh, it only works 80%, but it's great.
Speaker 2:I don't have to do the grunt work and talk to people. I don't need to deal with juniors and mid level engineers anymore. So, yeah, this is all true. You're working with with a lot of companies. Right?
Speaker 2:You have clients in Europe. You have clients in The US. Have you noticed a shift in expectations? Based on my experience, it comes in two flavors. One one shift I can see is that the expectation is you can do more and faster to to the extent where everything should be as simple as a prompt, and it almost never is.
Speaker 2:But, you know, that's that's the sentiment. And the other shift is that probably your team doesn't get enough credit because, hey, AI has done it. You know, those 30%, 40%, it wasn't done by AI. You don't get the credit. I feel that a lot of managers can relate to this because as managers, we never get the credit for success.
Speaker 2:We always get blamed, but that's probably a separate topic.
Speaker 1:This this question is honestly, in the last one year, I have had maybe one fourth of my customers come back who have been tech savvy to the point to telling me that, you know, okay. So so many number of hours, but, you know, are you using AI for it? What are you using? They don't because we focus on small and medium businesses. They don't really care what you use as long as you give it to them fast and you give it to them cheap.
Speaker 1:Right? So they don't really care. So there have been customers who've been asking, but not a lot. But the thing is that they do expect it to be cheaper now or, you know, at least faster deliver. So if if they think think that something is expensive or taking a lot of time, they do question, hey.
Speaker 1:Aren't you using air for this? You know, it should have been faster. Even though I I feel that they have no idea, it's just picking something from the air to say, okay. How much more can I, you know, get the cost down or get the timeline down to be able to use AI as a as a keyword to, you know, a bargain? It's it's a bargain point.
Speaker 1:You know? So so that I think that is that is what my experience has been with small and medium. But when we talk about enterprise or corporate customers, absolutely. They they have an understanding. You know, when I say we do legacy systems or logs logistics industry and we do ongoing maintenance, they they want so we have a lot of dedicated team members also sitting, you know, working with LA, working with Zurich.
Speaker 1:So they definitely, you know, are using a lot of AI. When when we're building bigger systems, client is also tech savvy, and they do question that. But other than that, small and medium clients, they don't care as long as you give it to them fast if you as long as you give it to them cheaper. And that's that's all they care about mostly. So I feel that in the small and medium industry, this is still coming.
Speaker 1:They do use it as a, but they don't really understand it. But within our own team, like I said, we do have to sort of, like, you know, cut the the cost down and make sure that we use AI to be able to generate. I don't wanna I don't wanna sort of, like, add one thing. You know? We keep talking about using AI to generate code.
Speaker 1:It's not that simple. Right? You you're you've been an engineer. I I've been an engineer. We understand that it's not that simple when people say when we say we're using AI.
Speaker 1:A lot of the time, what happens is that you're asking for a, and it'll give you an a minus. So if you're asking for an a, it'll give you a b. Sometimes they'll give you an a plus. Right? So a lot of the time right now, my seniors tell me that a lot of the time that we're spending so, you know, I'm not happy with with the amount of productivity or go going higher.
Speaker 1:I want more. Right? Being a business owner, that's what you want. Right? You want you want to be more profitable.
Speaker 1:You want more productivity. If you're using AI tools, why aren't we getting the amount of productivity that, you know, we should be getting? And these conversations we have very actively with our p six team and our seniors and all of that. So one of the things that we need to keep in mind and for our audience is that a lot of the times, AI will produce something that's absolutely amazing, and it, you know, saves you, like, 70% of your time. A lot of the time, it it you'll be wasting 30 more.
Speaker 1:So something that was hundred hours, you'd end up doing in a hundred and thirty hours because you were wasting thirty hours on actually looking at code that was crappy because you didn't you did not you you do as a senior, you never want to put in a plug and play code that you're not comfortable that you haven't seen, that you're not comfortable with because you don't know how it's going to impact the rest of the application. So you're definitely going to read it. You're going to audit it. And when you audit it, sometimes, you know, it's surprising, and you're like, wow. And sometimes you're like, what?
Speaker 1:And, you know, like and then you say, oh, this is just a waste of my time. I'm gonna scrap it, and I'm gonna do it myself. So there are areas that are AI strength, there are areas that are that are not its strength that you learn as you go.
Speaker 2:Do you notice this paradox where, on one hand, your you, and maybe your customers, maybe your p six team, expect AI to to increase productivity dramatically. Right? Because you invest a lot of time, you invest a lot of money. But on the other hand, you understand that it's still an emerging experimental technology, that it's nondeterministic, as you said. Right?
Speaker 2:It doesn't produce consistent results. And also, AI is not accountable by itself. Somebody else, like the person who just enters the prompt, is accountable for the result. And, of course, you can create a lot of tests. You can create guardrails almost like, you can create all kinds of agentic, non agentic processes around the LLMs and coding agents, but this has to be done.
Speaker 2:And the more you build them, the more rigid they become, and the the more maintenance they require just like processing it just in a human team. So do do you feel this paradox? Because you're supposed to have immediate productivity boost because AI is so amazing, but you don't have them right away. But you can't be skeptical and complain. You have to keep doing this because this is an emerging technology.
Speaker 2:Everybody expects you to use it. It will be amazing. It will be great a year from now, probably, but you're kinda stuck right now. That that's how I feel for sure.
Speaker 1:Not kinda stuck. I won't say kinda stuck, but you have to understand that you will grow organically in this. And you can't say, okay. You know what? We got you the tools.
Speaker 1:We're paying for it. I wanna see productive. Like I said, I had that conversation, and this is what I got. And I was like, yeah. Makes complete sense.
Speaker 1:Right? So Mhmm. I mean, you got some use cases out of it where somebody said, you know what? I I usually use AI to generate code. I got this task to actually audit code, to review code, because you also have teams who review code.
Speaker 1:Right? So to review code using AI. So what I did and this is just a one of the use cases, and I think that'll answer your question nicely, is that what I did is I reviewed code manually because I didn't trust AI to actually go and review code because AI generated the code, and I'm I'm wanting AI to go and review the code. So I reviewed it manually, and I found one major issue. Then I gave the same port to, you know, AI to review, and it found four issues.
Speaker 1:And one of those was the same one that I found. And then there was another issue that AI found, which was good to have, wasn't a must, but I liked it. Mhmm. Right? But then the other two that it found, and I spent time reviewing that and wasted time because that was crap.
Speaker 1:Mhmm. So what did I get out of it? I got learning that, yes, you can trust AI only so much, but not absolutely. Right? And then but but then you also have to understand, Alex, that, you know, last year, same time, we were more skeptical.
Speaker 1:Now we are not that skeptical. So it is improving. So we have to be patient, and we have to keep adapting it so that we don't get left behind in the race. And like I said, in 2025, I didn't have any expectations. I said, just go and play.
Speaker 1:And, you know, and and let's see what happens by the end of '20. Now that I see that, oh, we did improve productivity. We didn't grow the team, but we grew the revenue. So what happened? Right?
Speaker 1:So now I I I want more. Now I want to shrink the team a little bit and focus more on AI. So, you know, you have to take baby steps with it, but and be patient with it. That's that's that's what I think.
Speaker 2:Yeah. Yeah. That makes total sense. How do you convince your team to to change? I think change management is, like, a huge topic in itself, But with such a large organization, my assumption is that you can't mandate everyone to change and adopt.
Speaker 2:And people have the same challenge. Each individual has same challenges as the whole organization has. You have to do things you committed to today, and you also have to learn and adapt ever changing technology. And the the pace, it increased dramatically over the past couple years. How do you convince the team?
Speaker 2:How do you allocate time?
Speaker 1:I I think human nature is such that change is inevitable, but change is scary. So it is very scary. So I think it's not just about anything that comes in. So it's like I said, when when when this came so we gave them the first year as a as a playground to play with, and now we are pulling back on resources. We're basically, you know, shrinking the team and saying, okay.
Speaker 1:Hey. This is what it is. So now you're challenging them. You're you're you're making them step up. Because if you don't, some of them will step up organically, others will not.
Speaker 1:So apart from the fact that you are you are making the environment, you know, feasible for them to be able to adapt with, you know, training, so we have we have regular tech talks in the company, and most of them are around agentic, around prop engineering, and all of that recently. And then also also security is a is a huge risk of security. And then, you know, also challenging them with saying, hey. This is what it is. So, you know, for those, you know, bad benches or those low liars, they have to step up or they have to step out.
Speaker 1:So I think both.
Speaker 2:You've been doing this for many years. Right? And in two decades, you've been doing this. There must be some parallel you can draw with what's currently happening with the industry. Maybe it was big data back in the day.
Speaker 2:Maybe it was blockchain. Based on your experience, what comes to your mind as an analogy or or parallel, if anything?
Speaker 1:I think blockchain. Especially in twenty sixteen, seventeen, I was I I actually did a certification. I came to The US. I did a certification and brought it back home, and then, you know, everything was about blockchain. I don't know how it happens in other parts of the world, but, you know, the academia in Pakistan, they they go crazy about any new buzzword.
Speaker 1:They started degrees in blockchain and stuff like that, and, you know, people were and it it seemed like, you know, everything was going to go decentralized Mhmm. By 2026, 2027. And then, you know, in 2020 so it's like 2016, 2017. And then in twenty eighteen, nineteen, the bubble burst. So which is why I was also skeptical about AI, honestly.
Speaker 1:I was like, yeah. Yeah. It's gonna burst. But, see, it didn't. And, you know, this is this is like going back to when Internet came or, you know, you know, that revolution came Mhmm.
Speaker 1:When when mobile phone phones came. So, you know, things change if these evolutions happen, and then humans learn to evolve around it. Humans humans grow to be able to adapt and to be better than that. So the human race, I feel, is always going to be top of the line. Whatever I mean, something new comes, there is going to be chaos initially, but eventually, the the the the dust will settle, and there will be some sanity to it.
Speaker 1:So I think right now now AI dust is sort of, like, settling in just a little bit with the with with you know, when I say that, I mean that at least people are now more accepting of it and not closing their eyes to it that that, you know, this is here to stay. This is not another blockchain bubble. This is going to this is here to stay. So I think sanity is prevailing, and it'll prevail more in 2026, I'm hoping, and then people will start to adapt. And that I I I still hope and I pray that the human race is going to come up on top of it.
Speaker 1:People will have to learn and unlearn and relearn because that's that's the survival game right now.
Speaker 2:That wraps it up nicely and actually aligns with what I think as well. I'm also optimistic, and I'm glad to hear that you are as well. Let's live through this and and see what happens.
Speaker 1:No. Let's live through this and let's grow. Let's live through this and let's grow. That's that's the that's that's where that's that's the amount of optimism we should have.
Speaker 2:Absolutely. Shamim, thank you so much. It was my pleasure hosting you today.