Adam Kamel

Built the document intelligence and financial RAG platform behind a restructuring analytics product

• Cut document parsing cost 4x by replacing a paid service with a tuned open-source pipeline, saving $15K+
• Scaled ingestion to 100K+ financial filings on queue-based cloud compute
• Cut AI extraction hallucinations by 95% with deterministic validation checks

Leading WorldFold, a project exploring world models and reinforcement learning for robotic cloth manipulation

• Designed the staged, simulation-first roadmap from MuJoCo environments and PPO baselines through world-model experiments to sim-to-real deployment
• Established reproducibility standards with seeded trials, multi-episode evaluation, and random and scripted baselines
• Built the team's operating structure across recruiting, onboarding, research reading, and weekly technical delivery

Building an unsupervised pipeline to detect hidden market regimes in noisy financial time series data.

• Implemented regime discovery using K-Means baselines and time-aware Hidden Markov Models to infer latent market states
• Designed HMM state and transition modeling to capture regime persistence and shifts without labeled data
• Built regime-annotated price history visualizations to validate inferred states against major market events

Developed AI agent integrations and improved natural language processing systems

• Improved accuracy of enpoint search by 28%
• Developed 30+ service endpoint integrations with FastAPI
• Created comprehensive test suites using PyTest