At Handshake AI, I designed benchmark tasks and multi-tier scoring rubrics for Project Octahedron, and built gold-standard search trajectories for Project Seal India. I also developed command-line and code-execution evaluation tasks for Project Dynamo and audited multi-modal datasets to identify model reasoning failures.
At XDC Network, I direct QA and review workflows across a 17-member team, calibrate reviewers with structured feedback, and verify technical claims against primary sources, including RWA tokenization and ISO 20022 compliance standards. My work includes diagnosing hallucinated regulations and establishing verification checkpoints for complex commercial compliance evaluations.
