01 / THE PROBLEM
The problem
A stack of papers. A finite amount of time.
02 / WHAT I BUILT
What I built
AI-assisted summaries and insight extraction for academic papers.
03 / THE HARD PART
The hard part
A useful summary has to preserve the substance of the paper, not just shorten it. The central constraint is keeping generated insights grounded in the source.
04 / ARCHITECTURE
Architecture
A React interface connects to a Python / FastAPI backend, with language models and NLP supporting paper analysis.
05 / RESULT
Result
Research papers, made easier to explore.
06 / WHAT I’D DO DIFFERENTLY
What I’d do differently
The next useful evaluation would compare extracted claims with the original passages, including a way to flag uncertainty.