Mati Roy

Netholabs is a Whole Brain Emulation research company dedicated to preserving human agency in the age of AI. Backed by Foresight Institute, the Swiss National Science Foundation, and the NVIDIA Foresight Program, with research programs at the University of Sussex and Imperial College London. As Chief Product & Data Officer, I lead the data and product roadmap: turning Netholabs’ functional-emulation research into scalable datasets, model capabilities, product strategy, and early commercial partnerships. We're hiring ML engineers. If you're interested, please DM me!

AIChamp creates long-horizon tool-use datasets in realistic expert environments across domains including finance and academia. As CTO, I led the technical architecture, expert-hiring platform, labeling platform, and end-to-end data production workflows.

• Oversaw human data vendor operations • Built and led the coding human-in-the-loop team • Led the human data program shaping model alignment and conversational style

• TPM for post-training human data powering ChatGPT’s major capability improvements across instruction following, writing quality, multimodality, ChatGPT-specific tools (such as memory and code interpreter), general tool use, expert evals, and superalignment work • Partnered with 30+ researchers and teams across product, engineering, legal, and vendors • Built synthetic prompt generation pipelines to improve data coverage • Implemented evaluation scripts using external benchmark datasets; contributed to published alignment research (“The Alignment Problem from a Deep Learning Perspective”)

Founded and scaled a ~20-person distributed services firm delivering research and lead-generation support

Led hiring, operations, and infrastructure modernization for an emergency medicine research organization focused on deep hypothermia

To learn more about CCR's research and history: • https://biostasis.substack.com/p/a-man-for-all-ages-dr-steven-b-harris • https://www.researchgate.net/profile/Steven-Harris-14

Built data analysis and internal tooling (Python, R, MySQL)

• Ran large-scale factored cognition experiments with ~50 participants to stress-test debate-style approaches to open-ended reasoning • Owned end-to-end operations: onboarding, experiment design, execution, and iteration on platform robustness and incentives • Collaborated with research partners on evaluation methodology and experimental strategy Note: I joint as an experiment participant in July and got promoted in August. Our research project: https://ought.org/presentations/delegating-cognitive-work-2019-06

• Front-end owner for a Customer Interaction Management platform (React) • Sole developer of an internal Survey Engine web app (React, Node, SQL)

• Led frontend development of a next-gen electronic medical record platform integrating with hospital and lab systems (React/TypeScript, .NET, SQL, HL7) • Built internal reporting dashboards (ASP.NET/iTextSharp; BI tools: Tableau and SiSense) • Compared frameworks: ReactJS, AngularJS 1, BackboneJS and RequireJS • Scraped websites using PhantomJS in C# • Database maintenance • Established a technical support system for the support team Website: https://www.medfarsolutions.com/en_ca/myle/ Video of the app: https://www.youtube.com/watch?v=ORWof0sjTAA

• Creating a directory of robots with the Bioloid robot platform • Creating a fast prototyping method using this platform • Participating in the Canadian Undergraduate Physics Conference in Saskatchewan • Making slide shows to document the work from our laboratory • Developing a robotic arm on a turntable that controls its speed and acceleration, in C • Designed and led robotics workshops for students using LEGO Mindstorms