Tag: cmu

Privacy-preserving credit card fraud detection that runs ML inference directly on encrypted transaction data using TenSEAL (CKKS). Benchmarks 8 ML models under homomorphic encryption to quantify the accuracy/latency tradeoff.

A usability-driven redesign of picoCTF, built through iterative user research, heuristic evaluation, and formal usability testing as the final project for CMU 17-734 — Usable Security & Privacy.

RAG-powered threat analysis dashboard that combines a 3-stage LLM pipeline with CVE lookup, MITRE ATT&CK mapping, and static file analysis (YARA). Built as the final project for CMU 17-630 — Prompt Engineering.