Started my career in NLP research developing custom transformer based model. Then moved on to deploy big data pipelines at scale for banks. Then developed production-ready LLM workflows for insurance and for my own company: SubstantAI. Always designed and built solution end-to-end incl. cloud architecture, devops, full-stack web development...
- mcpadapt: mcpadapt bring 650+ mcp servers tools to any agentic framework seamlessly, it is merged in huggingface/smolagents and CrewAI and how they support MCP today.
- OAuth Claude Code Action: a fork of Anthropic's github "@claude" action which enable any paying users of Claude to use "@claude" at no additional cost.
- iodmcp: iod is an organisation with a few repo behind the iod.ai website. We facilitate (even more) access to tools and data to LLM via the MCP protocol.
- [2025.06] - HF x Anthropic AgentHack Hackathon Finalist: built the first ever Apple Health MCP (leveraging HF spaces to remain private) and then created a multi-agent deep research system on top your data and public medical paper data. #MCP #Smolagents #HF
- [2025.06] - OpenAI Viva LeHack Hackathon Finalist: built a conference guide helping attendees search for relevant events based on the conference agenda and their linkedin profile. It also used google calendar MCP to automatically add events in your agenda and handle conflicts. #MCP #OpenAIAgentsSDK
- [2024.07] - Mistral Finetuning Hackathon Winner: built a multi-agent system with fine-tuned models grounding swiss legal question on top of swiss law. #AutoGen #Mistral
- [2024.07] - Kaggle Competition Solo Bronze medal: fine-tuned and quantized various models including gemma 2 9B on LMSYS Chatbot Arena human preference prediction task.
- [2022] - Darts: User-friendly modern machine learning for time series (JMLR) - developed the backtesting approach of darts a time series forecasting framework.
- [2020] - Fast cross-domain data augmentation through neural sentence editing - implemented from scratch a novel transformer model combining VAE to perform cross-domain data augmentation via semantic preserving edits.
- Email: [email protected]
- LinkedIn: https://www.linkedin.com/in/guillaumeraille/
Feel free to explore my repositories and reach out for colaborations!