Hi, I’m the CEO and Founder of Cowork AI. My current focus is building Agentic AI and LLM-driven trading systems for US equities and cryptocurrencies.
Previously, I was an Entrepreneur in Residence at Hashed, where I explored and developed web3 projects. Prior to my startup journey, I spent five years as a Software Engineer at Waymo (formerly the Google Self-Driving Car Project), deploying vision-based machine learning systems directly to the autonomous fleet.
š¼ Experience #
Cowork AI, Inc. #
CEO & Founder (Nov 2022 ā Present) I founded Cowork AI to bridge the gap between foundation models and live financial markets.
- Project Hayes: Deployed an autonomous Agentic Trading system actively powering Cowork AI’s proprietary trading. It monetizes 2nd and 3rd-order market effects using Knowledge Graphs, RAG, and Chain-of-Thought reasoning.
- Fred MCP Server: Launched an integration layer that seamlessly connects foundation models to real-time financial feeds, technical indicators, and alternative data sources (e.g., Reddit, Truth Social).
- Partnership: Supplied the Fred MCP Server to Kakaopay Securities, powering Korea’s first AI mobile trading feature that autonomously researches and explains the causes behind sudden stock price movements.
- Fred RAG Agent: Released an enterprise-grade, on-premise Agentic RAG solution tailored for financial and accounting firms.
Hashed #
Entrepreneur in Residence (Oct 2021 ā Mar 2022)
- Pass the Baton: Launched an ERC-721 NFT art project on Ethereum designed as a sustainable platform for charitable giving (baton.art).
- Operated as a two-person development team, personally architecting and writing half of the production Solidity smart contracts.
Waymo LLC #
Software Engineer (May 2016 ā Jun 2021)
- Deployed vision-based detection systems to the self-driving fleet, unblocking new driving areas by classifying unique, dynamic traffic signals.
- Patented Technology: Co-invented a machine learning system for autonomous vehicles to detect, classify, and respond to unfamiliar traffic signs in real-time (US Patent 10,928,828) (Archive).
Nexon Corp. #
Software Engineer (Sep 2006 ā Apr 2009)
- Full-stack developer for MapleStory, a blockbuster 2D side-scrolling MMORPG. Developed a C++/DirectX 3D game client and an IOCP-based highly-concurrent game server.
š ļø Technical Skills #
- Agentic AI & Protocols: Multi-Agent Systems (MAS), Model Context Protocol (MCP), Google ADK, Gemini CLI, RAG, Knowledge Graphs.
- ML & Infrastructure: vLLM, llama.cpp, PyTorch, TensorFlow, Vertex AI, Vector Databases (pgvector, SQLite-vector).
- Software & Cloud: Go, Python, C++, Solidity, gRPC, Docker, Nix, PostgreSQL, GCP (Cloud Run, GCE), HTMX, Alpine.js.
š Education #
- University of California, Berkeley
- Ph.D. Candidate (On Leave) in Computer Science (Aug 2012 ā May 2016)
- M.S. in Computer Science (May 2016) | Thesis: View-Dependent Adaptive Cloth Simulation (Archive)
- Seoul National University
- B.S. in Computer Science and Engineering (Feb 2012)
š Selected Publications #
- View-Dependent Adaptive Cloth Simulation with Buckling Compensation
Woojong Koh, Rahul Narain, and James F. O’Brien. IEEE Transactions on Visualization and Computer Graphics (TVCG), October 2015. - View-Dependent Adaptive Cloth Simulation
Woojong Koh, Rahul Narain, and James F. O’Brien. ACM SIGGRAPH/Eurographics Symposium on Computer Animation (SCA), July 2014. - View-Dependent Adaptive Cloth Simulation (Archive)
Woojong Koh. Master’s Thesis, EECS Department, University of California, Berkeley, Technical Report No. UCB/EECS-2016-5, January 2016. - Near-exhaustive Precomputation of Secondary Cloth Effects
Doyub Kim, Woojong Koh, Rahul Narain, Kayvon Fatahalian, Adrien Treuille, and James F. O’Brien. ACM SIGGRAPH 2013, July 2013.