Problem

CS students can't tell which roles fit their skills, or what to learn next to get there. This started as Roshni's Graduate Assistant work with Binghamton's Watson Career Services and grew into a full platform.

Approach

Python/FastAPI and GraphQL services on AWS EC2 with a React/TypeScript frontend. BERT sentence-transformers semantically match user skills against O*NET occupational data plus custom modern roles (AI Engineer, DevOps, Cloud Architect). A skill-gap engine scores what's missing for a target role and React Flow renders the roadmap. RAG-based resume analysis with PDF parsing, plain-English LLM explanations for every recommendation, and YouTube API course suggestions for each gap.

Results

205 role pathways with matching, gap analysis and personalised learning paths. A 17-case pytest suite across 10 modules covers retrieval quality, scoring accuracy, auth persistence and security validation, plus Jest frontend tests and a smoke test against Docker Compose.