Chris Nielson
builds AI products
Software engineer in Brooklyn, NY. I build production AI systems: agents that work with real tools, data, and permissions, plus the knowledge graphs and context engineering that keep them grounded.
After hours I run open-weight models on my own hardware to see how far small, local models can go when the context is right.
selected projects
AMATS
Adaptive multi-asset trading system for equities and crypto. 25+ specialized agents debate each decision before a risk guardian approves, modifies, or vetoes it. Paper and live execution, backtesting, risk controls.
Local inference server for Apple Silicon. Hot-swaps MLX models (LLM, vision, embeddings, TTS, STT) behind an OpenAI-compatible API.
Entity Resolution Workbench
Auditable system for resolving 76k+ procurement vendors against 196k+ finance records: multi-pass blocking, interpretable scoring, one-to-one resolution, review queues, and knowledge-graph audit trails.
Multi-engine risk analysis comparing QUIC and Acadia outputs across FX swaps and forwards: PFE metrics, decision-tree analysis, interactive visualizations, and LLM-generated explanations.
writing
- GraphRAG as Scoped Context for Production Agents
- Semantic Models for Databases
- Production AI Agents in Enterprise Systems