#01

AI Candidate Evaluation System

Team · 5 engineers
Sole full-stack engineer on a 5-engineer team

Evaluates a software candidate in under 2 minutes by pulling their public GitHub repos (about 30 on average) and turning them into a structured recruiter report.

key metrics · highlights
<2 minper candidate
~30repos analyzed on average
5engineers on the team
mission briefing · details
  • >CodeBERT embeddings from repo source code, with analysis routed through the Gemini API.
  • >Reports cover code quality scores, language proficiency and contribution patterns.
  • >Job-description requirements are indexed as retrieval queries against the candidate's embedded code, giving evidence-backed skill-gap comparisons.
  • >Built the FastAPI backend, the LangChain orchestration layer and the React recruiter dashboard.
tech stack
PythonFastAPILangChainCodeBERTGemini APIGitHub APIReact
year2024