I'm Milan — a Master's student in Artificial Intelligence & Machine Learning at Drexel University. I work across the full stack of applied AI: training and evaluating language models, building LLM-powered products, and shipping them to production. I'm most interested in LLMs, agentic systems, and making ML systems actually ship.

Languages

ML / AI

Backend & Web

Cloud & DevOps

Tools

• GapScope — Engineering Knowledge-Gap Analyzer — A deployed LLM-powered career-gap analyzer (Next.js + Claude). Mines evidence from public GitHub repos, grounds it with a structured LLM pass, and returns a trajectory-aware learning path, project gaps, and a positioning read. Built with hallucination guardrails — HTTP-verified resource links and validated graph-shaped output.
• Political Bias Transfer in LLM-Generated Training Data — A multi-axis audit of whether an LLM's political lean transfers into classifiers trained on its synthetic output. 189 fine-tuning runs across 7 generators and 3 architectures, with word-level mechanism analysis.
• Neuflo Learn Backend — A FastAPI + PostgreSQL backend powering an adaptive NEET exam-prep platform: closed-loop test generation, proficiency tracking, and per-chapter adaptive feedback.