AI Systems · Routing
Enterprise AI Workload Intelligence
Routing AI workloads across cost, quality, reliability, latency, sparse-history backoff, and confidence-aware decisioning.
Each project has its own article. The article explains the problem, architecture, evaluation, tradeoffs, and what changed after testing it.
Routing AI workloads across cost, quality, reliability, latency, sparse-history backoff, and confidence-aware decisioning.
A DeFi risk and stress-testing lab covering shock ladders, correlated Monte Carlo simulation, calibration, and liquidation threshold sensitivity.
A browser-based distributed compute experiment for discovering peers, negotiating direct connections, and routing work across available devices.
A multi-stage classification system for standardizing inconsistent legal charge text across jurisdictions with retrieval, ensembles, confidence gates, and LLM review.