At FastCode AI, I designed authorization and PostgreSQL systems for a legal corpus, closing data-exposure gaps and fixing integrity and reliability defects. I scaled ingestion to 105 courts and 1.6M cases, and built a Qdrant retrieval pipeline validated across 338 checks.
I also developed legal case-readiness rules that recovered a 4.82M-case gap, and built a DINOv2 + CenterNet vision pipeline that achieved 60.8% F1. As an ML Intern at FastCode AI, I worked on these production systems from January to August 2026.
Earlier, I developed workflow automation services at Revscale AI and machine-learning pipelines at RAPID, PES University. My projects include an AI investment system applied in live trading and papAIrus, a voice-first platform that turns journaling into a personal knowledge graph.
