Multi-Agent RL Market-Making Engine (MARL)latest
Trained dual-swarms of AI agents to compete and cooperate in a simulated stock exchange, written from scratch with PyTorch and formal mathematical specifications.
Trained dual-swarms of AI agents to compete and cooperate in a simulated stock exchange, written from scratch with PyTorch and formal mathematical specifications.
A machine-learning engine that decides which ad to show and when — written entirely in NumPy from scratch, with 23 verified mathematical spec checks.
Built a declarative multi-agent AI orchestration engine with 300+ tests and hard safety guardrails, including parallel execution contributions to the open-source Goose framework.
An embedded property graph and retrieval pipeline indexing 141 sessions of AI coding history with BM25 FTS and semantic search.
An AI-native research operations backend enforcing formal process invariants via FastAPI, PostgreSQL/pgvector, and NATS.
A systemd monitoring daemon that polls social feeds, evaluates sentiment using Gemini 3.5 Flash, and checks impacts against live global market contexts.
A rollup of minor utilities, infrastructure projects, and side experiments — from in-memory PDF merging to multi-cloud orchestration and MCP bridges.
Trained dual-swarms of AI agents to compete and cooperate in a simulated stock exchange, written from scratch with PyTorch and formal mathematical specifications.
A machine-learning engine that decides which ad to show and when — written entirely in NumPy from scratch, with 23 verified mathematical spec checks.
Built a declarative multi-agent AI orchestration engine with 300+ tests and hard safety guardrails, including parallel execution contributions to the open-source Goose framework.
An embedded property graph and retrieval pipeline indexing 141 sessions of AI coding history with BM25 FTS and semantic search.
An AI-native research operations backend enforcing formal process invariants via FastAPI, PostgreSQL/pgvector, and NATS.
A systemd monitoring daemon that polls social feeds, evaluates sentiment using Gemini 3.5 Flash, and checks impacts against live global market contexts.
A rollup of minor utilities, infrastructure projects, and side experiments — from in-memory PDF merging to multi-cloud orchestration and MCP bridges.