Zaid Rahim / Lahore, Pakistan
I build LLM systems that reason over real-world data.
AI/ML engineer focused on production RAG pipelines, LLM fine-tuning, and the Python backends that serve them — from attention math written by hand to models shipped behind a FastAPI endpoint.
Where I go deep
Six areas I keep returning to — the ones I understand from the math up, not just the API down.
Projects, as case studies
Real systems with real constraints. Expand any project for the problem it solves and what I built.
Built in the open
Most of my work lives on GitHub. Here's how it breaks down, plus a few repos beyond the case studies above.
Focus by domain
Public repositories, grouped by what they are. Star and fork counts aren't shown — they weren't part of the exported data, and I'd rather show nothing than invent numbers.
View full profileMore repositories
Experience & education
Technical skills
Grouped by what they're for — every item here is something I've actually shipped or built with.
I got into AI by refusing to treat it as a black box.
I'm a final-year Computer Science student at FAST NUCES, and most of what I know about machine learning I learned by rebuilding it. Before I trusted PyTorch's autograd, I wrote backpropagation by hand in NumPy. Before I called a Transformer, I implemented attention, positional encoding, and residual streams one module at a time. That habit — build it to understand it — is how I work.
These days I build systems people can actually use: a RAG assistant that grounds legal answers in 5,000+ real judgments, a 3B model fine-tuned with QLoRA on a single free GPU, semantic search over thousands of documents. I care about the parts that make AI trustworthy — retrieval that cites its sources, inference that can run locally and privately, and backends documented well enough for someone else to pick up.
I've also taught operating systems to 30+ students and led engineering teams of up to ten, which taught me that the clearest explanation usually wins. I'm looking for AI/ML engineering work where I can keep shipping real systems.
Let's build something.
I'm open to AI/ML engineering roles and interesting collaborations. The fastest way to reach me is email — or send a note below.