Six years in, I've stopped thinking of "data science" and "AI engineering" as separate jobs. A model that never leaves a notebook doesn't help anyone — so most of my time goes into the unglamorous middle: turning a promising experiment into something that survives real traffic, real data drift, and real users who don't read the docs.
Lately that means a lot of agentic systems — LLMs that don't just answer, but plan, call tools, and know when to ask for help instead of guessing. I also spend a good chunk of time on the teaching side of this: guest lecturing on Agentic AI, Conversational AI, and LLMs, because explaining a system out loud is the fastest way to find the part you don't actually understand yet.
|
🧪 Machine Learning & AI
|
📐 Data Science Practice
|
⚙️ Production-minded AI
|
🪴 What I'm building — KUWT
KUWT is an AI-powered content generation automation platform — built to keep technology enthusiasts and learners current with what's actually happening in AI, without the noise.
🐇 Outside the day job — bunnybethinking
I run bunnybethinking, an AI/ML content brand built around a simple idea: be the mentor I never had when I was starting out. Most AI content online either talks down to beginners or talks past them — I'm trying to build something in between, aimed at people who are curious, a little overwhelmed, and tired of feeling like an outsider in this space.
It shows up as fundamentals breakdowns, weekly paper reads, "what I tested this week" experiments, and painpoints pulled straight from what people are actually asking on Reddit — not a course funnel, just a village of people figuring this out together. No-filter, comfortable being wrong, allergic to hype.
Most of my day-to-day is applied — shipping systems, not training foundation models. But I've been pulling back toward the fundamentals lately: how LLMs are actually pretrained, the architecture and optimization choices underneath the abstractions I usually work above. It's partly curiosity, partly the belief that you can't reason well about agentic systems, fine-tuning, or failure modes if you've only ever worked with a model as a black box behind an API.
I guest lecture on Agentic AI, Conversational AI, and LLMs at BITS Pilani's Work Integrated Learning Programmes. I like teaching this stuff for a selfish reason too — explaining a system out loud is the fastest way to find the part you don't actually understand yet, so it feeds back into how I build.
