Andrew Metcalf

These projects look different on the surface but share a common foundation: building complex systems that perceive, remember, and act in the real world. That is also the direction of my PhD research, cognitive AI systems with persistent memory and identity.

Ebon Crucible is a real-time combat system with 63 interlocking abilities, crowd control chains, and an economy that has to stay balanced. AlphaTelemetry Lab runs autonomous trading strategies through staged validation across three asset classes. CacheX turns a single photo into a live portfolio of resale-priced assets and keeps it current over time. Ajax Studio orchestrates multiple AI models into creative output with persistent identity. VoiceGuard detects synthetic voices in the wild. Each requires a different stack, but the same discipline: ship it, run it in production, and make it work without hand-holding.

The common thread is memory and grounding. An AI that forgets cannot maintain an identity. A trading system that ignores reality blows up. A valuation tool that uses stale data lies. The PhD direction grew directly out of building these systems: figuring out how machines should remember, what they should forget, and how they should stay connected to the real world they are acting in. The breadth is the point. Building across domains reveals patterns that specialization misses.