Interviewing Aniruth

Curiosity. Whenever I come across something I don't know, I try to learn it — not to become an expert, but to have enough fluency to understand what's going on. So if something breaks, I can fix it myself instead of depending on someone else. This isn't just technical. I try to do this in general too.

Back in 2020, my aunt worked at a small company. Part of her job was consolidating hundreds of client invoices into a report, by hand, every time. It used to eat up days of her time. I built her a tool: parsed the PDFs to text, used regex to pull out the fields, and generated the consolidated report using Jinja templates in her company's format. This was pre-LLM, so this kind of thing actually took real work.

The harder part was giving it to someone non-technical. I couldn't hand her a script and a terminal. So I built a GUI in tkinter — file/folder pickers, client selection, the works — and packaged it into an executable with cxfreeze so she didn't have to install Python or any dependencies. I kept the regex rules in an external config file, so if an invoice format changed, she — or I — could fix it without rebuilding the whole app.

What used to take her 5-7 days took an hour. That's one of the things I'm most proud of building.

I started at TCS in a functional testing role, working on the team's WebdriverIO BDD framework. Within six months, with barely any handover, I was working at the level of the senior folks on it. I also worked on ML initiatives, a VR application, and a React Native app for client visits. I helped move the testing framework from sync to async and from JavaScript to TypeScript, and standardized it for the team. My work had real visibility on the client side — client leadership knew me and my work by name. But eventually it got repetitive and I stopped learning. So I moved into Java development at Wipro, since dev work was what I actually wanted. Deployment and infra work came as part of that role, not as a separate track. Same at HCLTech, where I now lead the effort moving our banking application from on-prem to GCP.

Honestly, I haven't run into real situations where I needed to write a sorting algorithm from scratch, so I might be rusty on it. But if a job actually needs that depth, I'll learn it. I have no resistance to learning new things — that's kind of the whole premise of who I am.

Generalist. I don't want to tie myself to one stack or language. Languages are tools for the problem in front of you — if all you have is a hammer, everything looks like a nail. I'd rather know what's actually best for the problem and use that.

Comfortable on both.

One where my work has real impact, and where people are treated like adults — decisions make logical sense, not arbitrary rules for their own sake. People should have the autonomy to do what's actually best rather than follow process for the sake of process.

Learning AI agents and the broader AI systems space — I find it genuinely interesting. Also working on my fitness, so don't be surprised if you catch me at the gym. More detail on my /now page.

Useful — it adds a layer of abstraction where you're not thinking about every line but looking at things from a higher level. But you still need to judge whether a piece of code is actually needed or correct. A human has to sign off, not just accept whatever the AI produces.