I'm an AI engineer deeply interested in language modelling — how models learn, represent, and generate language. I like getting into the internals: training loops, attention mechanisms, tokenization, inference pipelines — the things that make LLMs actually work.
Right now, I'm focused on building a reasoning model from scratch and optimizing model inferencing.
- Production-grade RAG applications — end-to-end retrieval-augmented generation systems designed for real-world reliability and scale
- Guardrails from scratch — custom input/output validation and safety layers for LLM-powered applications, built without leaning on off-the-shelf frameworks
- Agentic frameworks — multi-step, tool-using agent architectures that go beyond simple prompt-response patterns
- Reasoning model (in progress) — training a reasoning model from the ground up to understand how chain-of-thought and structured thinking emerge
- Inference optimization — working on making models faster and cheaper to serve at scale
AI / ML
LLM Frameworks & Tools
Backend
Frontend
Databases & Vector Stores
Other
