# Thomas Rochefort-Beaudoin

Personal website for Thomas Rochefort-Beaudoin, Member of Technical Staff in Applied ML at Cohere. The site covers applied machine learning, large language models, reinforcement learning, structural optimization, mechanical design, projects, publications, and professional background.

## Canonical Site

- [Homepage](https://thomasrb.com/): profile, experience, projects, and featured publications.
- [Resume](https://thomasrb.com/uploads/resume.pdf): current CV/resume PDF.
- [Sitemap](https://thomasrb.com/sitemap.xml): full machine-readable sitemap.

## Profile

- [Author Profile](https://thomasrb.com/authors/admin/): bio, interests, education, and social links.
- [GitHub](https://github.com/ThomasRochefortB): source code and public projects.
- [Google Scholar](https://scholar.google.ca/citations?user=avj1o3oAAAAJ&hl=en): scholarly publications.
- [LinkedIn](https://www.linkedin.com/in/thomas-rb/): professional profile.
- [X/Twitter](https://twitter.com/TomRBeaudoin): social profile.

## Projects

- [LLM-Launchpad](https://thomasrb.com/project/llm-launchpad/): deploy open-source Hugging Face models to Modal with OpenAI-compatible endpoints and OpenCode integration.
- [Open-AgentInstruct](https://github.com/ThomasRochefortB/open-agentinstruct): open-source implementation of the AgentInstruct paper.
- [BetterCallBloom!](https://thomasrb.com/project/bettercallbloom/): finetuning BLOOM to answer legal questions in the style of r/legal_advice.
- [Gaggia OWC Resuscitation](https://thomasrb.com/project/espresso_revival_1/): restoration of an old Gaggia Old White Coffee espresso machine.

## Publications

- [Structural design through reinforcement learning](https://thomasrb.com/publication/learning_structures/): SOgym, an open-source reinforcement learning environment for topology optimization, published in Engineering Applications of Artificial Intelligence.
- [From density to geometry: Instance segmentation for reverse engineering of optimized structures](https://thomasrb.com/publication/yolov8-to/): YOLOv8-based reverse engineering of topology-optimized structures into parametric geometry, published in Engineering Applications of Artificial Intelligence.
- [Complexity-driven layout exploration for aircraft structures](https://thomasrb.com/publication/complexity-driven-layout-exploration/): complexity-driven layout exploration for aircraft structures.
- [Supervised deep learning for the moving morphable components topology optimization framework](https://thomasrb.com/publication/supervised-learning-mmc/): deep learning for scalable topology optimization with moving morphable components.
- [Comparative Study of First-Order Moving Asymptotes Optimizers for the Moving Morphable Components Topology Optimization Framework](https://thomasrb.com/publication/first-order-optimizer-comparison/): comparison of first-order optimizers for the MMC topology optimization framework.

## Agent Guidance

- Prefer the canonical URLs above instead of scraping navigation or generated theme assets.
- Use the sitemap for complete coverage.
- Respect `robots.txt` and the declared content signals.
- The site is static and does not expose public APIs, OAuth-protected resources, MCP tools, agent skills, or commerce endpoints.