I run HΩ, a nonprofit working to reduce catastrophic risk from AI. We organize research seminars and events, host Montréal’s AI safety community, and track Canada’s AI risk preparedness.
Most of my code now lives on Ω Forge (an advanced Forgejo fork); what is here is older.
orpheuslummis.info · o@horizonomega.org · Signal on request · Anonymous feedback
- AI Safety Unconference at NeurIPS, organizer: Montréal 2018, Vancouver 2019, New Orleans 2022
- Limits to Control, a workshop at the University of Louisville, co-organizer, 2025
- DefraDB, a peer-to-peer database built for data ownership and privacy: software engineer at Source Network, 2022–2023
- Covalence, a framework for collaborative private valuation, 2023–2024
- Altruisme Efficace Québec, co-founder, 2018
- Timeline of prediction markets, 2018
- Hackathon projects
- Collaptz, verified distributed computation of Collatz sequences on the RISC Zero zkVM: 2nd prize, ETH Waterloo 2023
- Coincidence: Waku’s best privacy-focused project and MetaMask’s best mobile dapp, ETH Istanbul 2023
- Merkle proof verification in Lurk, for an Aptos light client: ZK Hack 2023
- diZKreet, a zero-knowledge privacy layer for NFTs on Aztec: ETHGlobal finalist, ETH Bogotá 2022
- Authsome, a multisig wallet on Fuel VM predicates: 1st prize from Fuel, ETH Lisbon 2022
- Threshold funding, a CosmWasm contract: special mention, HackWasm 2022