hi, hafeez here.
/ how i build
Find the smallest version that can prove itself.
I turn broad, ambiguous problems into focused workflows that can be built quickly, put in front of real work, and improved from evidence.
Make every step testable.
Capable AI is easier to trust when its work is visible. I design systems as smaller steps that can be evaluated, debugged, and made accurate independently.
Build around the domain, not just the model.
The useful system is usually bigger than the AI inside it. I combine domain knowledge, models, code, and existing tools so the result fits how people actually work.
Current toolkit:
- Agent systems: LangGraph, multi-agent orchestration, evaluation workflows
- AI-native development: Python, JavaScript/TypeScript, Vercel AI SDK
- Automation and integrations: n8n, CRM platforms, APIs
- Computer vision: PyTorch, OpenCV, YOLO, RT-DETR, ONNX Runtime
- Production ML: Docker, Replicate/Cog, CI/CD, inference pipelines
- Systems: Rust, AWS, Vercel
/ now
snapshots of what i am up to
LAST_UPDATE: September 17, 2026 • Bangalore
- Working on
AI systems, coffee, and Claude
Most days are a mix of building scalable AI systems and automations, with coffee within reach and Claude open somewhere nearby.
- Recent win
An ADHD-focused hackathon
My team won, and I built the backend, the AI layer, and a live assistant that was especially fun to bring to life.
See the project ↗ - Currently reading
East of Eden
John Steinbeck
- On repeat
Tame Impala
Enjoying the small craft of curating playlists instead of letting an algorithm choose everything.
- Training
Running and strength
Trying to become more consistent with both and make movement a dependable part of the week.
- Want to try
Bouldering
It has been on my mind for a while, so I plan to finally give it a proper try.
- Sidequesting
Magnus Carlsen, LIVE
Recently got to watch Magnus Carlsen play LIVE. One thing off the bucket list!
- Planning
More solo trips
Leaving a little more room this year for unfamiliar places and time on my own.