Remembering Tata Elxsi: How My First Corporate Job Shaped My Entire Engineering DNA

July 4th 2005. While a country was celebrating independence, I was celebrating a different kind of milestone: my first day in the corporate world at Tata Elxsi.

I walked into the office carrying a backpack full of computer science theory, a massive wave of imposter syndrome, and absolutely zero idea of what I’d actually be doing. The training programme was a daunting six-month long period and there was a looming two-year bond.

For a fresh graduate, it felt overwhelming. But looking back, over two decades later, as I now look at building and scaling Sudama Health, I realize that my entire framework for problem-solving was forged in those exact hallways.

The initial training was an intense mixed bag. I found myself yawning through basic C and C++ refreshers but completely captivated by Linux kernel internals, device drivers and Matlab.

When we were finally released into the real world of actual projects, I landed in a group tasked with building an image processing library entirely from scratch. Remembering my trusty copy of Computer Graphics by Foley et al., I felt a surge of confidence. I was ready.

My first assignment? Implement a simple image blur feature.

I wrote the code, deployed it to our state-of-the-art TigerSHARC DSP processor and ran it on a 1MB image.

It took 26 seconds.

In the modern world of instant gratification, 26 seconds is an eternity. But in that moment, I was ecstatic. The code worked.

I spent the next few hours optimizing, tweaking and pulling levers until I managed to compress that execution time down to 6 seconds. Feeling like an absolute engineering prodigy, I confidently walked into our client meeting.

Then came the reality check.

The client looked at the 6-second benchmark, looked back at me and calmly stated their target: 20 milliseconds.

I wasn't just off, I was off by several orders of magnitude.

What followed was a masterclass in first-principles thinking. My seniors didn't mock my failure, instead, they forced me to strip away all my high-level assumptions and look directly at the hardware. They introduced me to the complexities of Direct Memory Access (DMA) and hardware architecture.

I buried myself in research papers, hunting for faster algorithms. With a combination of algorithmic efficiency and code optimization, I managed to squeeze the performance down to just over 200 milliseconds.

Better, but still 10x away from the target.

"Let's write the whole thing in Assembly," my seniors suggested. Surely, dropping down to absolute bedrock metal would solve it.

We did. And the performance inexplicably worsened to over 300 milliseconds.

We were completely stumped. I began painstakingly debugging, counting CPU cycles and trying to find where our execution was bleeding out.

And then, we found it. The culprit was a single operation: the divide.

While a divide instruction existed in the assembly language, we discovered it translated into a heavy, clock-cycle-hogging software divide inside our specific processor architecture.

The fix? We stripped out the divide operation entirely, replacing it with bitwise shift operators and subtraction.

The bottleneck evaporated. We hit the 20-millisecond target.

That moment changed me. It was the exact flashpoint where I realized that when a problem looks impossible, you don't look for shinier tools, you dig deeper into the fundamentals until you find the hidden software divide choking the system. It’s the exact same first-principles lens I use today when trying to untangle complex systems in healthcare. I would not have the confidence I have today to tackle ambiguous, systemic problems had I not started at Tata Elxsi.

Tata Elxsi wasn't a perfect company, no company is. But in 2005, very few places in India were doing deep, foundational technical work of that caliber. Even today, they stand alone in many ways.

But their greatest achievement wasn’t just the tech, it was the ecosystem. They gathered an incredible concentration of brilliant minds. The cohort of fresh grads who stumbled through those hallways together trying to figure out corporate life became the people who are doing extraordinary things across the global tech landscape today.

More importantly, they became the bedrock of my personal life. To my lifelong friends  and the mentors who guided a naive CS grad, thank you. We built more than just libraries, we built a foundation that has lasted a lifetime.

With the embedded board at Tata Elxsi