About
Who I am and how I work.
I started in chemical engineering at IIT Kharagpur, then switched to software. The reason is simple: I wanted to build things I could see fail and understand why, not just design on paper. Six years later, I still chase that same itch.
I've worked across the stack — full-stack applications, backend services, infrastructure, distributed systems, and down to the systems level where you have to understand what the kernel is actually doing. But I don't stay in one layer. I'll drop to the database if that's where the bottleneck is, move to the systems layer if I need to understand what's actually happening, or climb back up to the application if the full picture matters. Right now I'm at Oblivious building LLM infrastructure and confidential compute tooling, but the core question is always the same: how do I make this observable, predictable, and trustworthy?
I also explore the edges on purpose — I've built vsock proxies and a Rust terminal editor — because that's where I understand what the layers above actually rest on. I read a lot, I code a lot, and I think about systems before I optimize them.
How I work
I care about correctness and clarity over shipping fast. I'll spend an extra day on fundamentals if it saves a month of firefighting later. I write code that's explicit, not clever, and I'd rather repeat myself than hide constraints inside abstractions that look neat but don't hold up.
I'm skeptical of tools and frameworks until I know what problem they actually solve — but when something works, I adopt it. I use Claude and other AI tools not because they're shiny, but because they let me move faster on the parts that don't need deep thought, and focus on the hard problems that do. I learn fast across different domains because I focus on principles, not just patterns. Ask me a question I haven't answered before, and I probably have a system for working through it.
Currently
At Oblivious building LLM infrastructure. I'm learning confidential computing, AI ops, and how to apply systems thinking to a new layer. The specifics change, but the work is always about making something reliable and observable that wasn't before.
Say hello
Want to talk infrastructure, LLM systems, or something else entirely? I actually read my email. Find me at asifiqbal7.iitkgp@gmail.com, or on GitHub.