01
Idea
Using LLMs as structured data augmentation systems
Notes on treating language models as constrained workflow components rather than generic chatbot layers.
Writing
Technical notes on architecture, research tradeoffs and product lessons.
Short, direct writeups on decisions that shaped the work: what was built, what failed, what improved, and what should happen next.
02
Draft
What I learned building a drug-discovery pipeline as an undergraduate
A practical reflection on architecture, validation, dataset bias, reinforcement learning and chemical representation choices.
03
Idea
From ambiguous problem to shipped product
Lessons from building startup software where technical decisions, sales conversations and product strategy collide.