Romain Zimmer — ML/AI Engineer
https://romainzimmer.com
Paris, France (open to relocation)

SUMMARY
ML/AI engineer with 7+ years building ML and AI systems and production-scale ML pipelines. Deep interest in AI algorithms and systems, model architectures, training methods, and infrastructure, alongside hands-on experience taking ML from research to deployment. Currently cofounder/CTO of Correl8.ai, a product observability platform for AI agents.

EXPERIENCE

Correl8.ai — Cofounder & CTO (Jan 2026 – Present, Paris, France)
- Cofounded and lead technology for correl8, a product observability platform that turns AI agent conversations into structured product signals (observations, sentiment, grouped issues).
- Designed agentic processing pipelines with vLLM inference: LLM-based ingestion (span classification, async issue digest, tag matching), Correl8 Assistant (in-app conversational agent), and background agents for autonomous analysis.
- Built MCP and REST integrations for third-party agents (LangChain, Pydantic AI, OpenAI Agents SDK) to post observations and run in-conversation research studies.
- Own full technical architecture and product direction as CTO.

Too Good To Go — Machine Learning Engineer (Dec 2022 – Dec 2025, Paris, France)
- Designed and built a discount optimization system from scratch for retailers (waste reduction, profit).
- Data pipelines with DBT and Snowflake; ML workflows with Metaflow.
- Predictive models for discount elasticity and a decision engine for optimal discounting.
- Built an internal experimentation framework for A/B tests of ML models; ran in-store tests with significant impact across retailers.

Freelance — MLOps Engineer (May 2022 – Dec 2022, Paris, France)
- Scalable ML infrastructure and CI/CD.
- Refactored Python codebase, integration tests, GCP to AWS Serverless migration.
- Triton Inference Server on AWS SageMaker; Slack bots for ML validation.

Sievable — Technical Co-Founder (Jul 2020 – Apr 2022, Paris, France)
- Inria Startup Studio funding; ML-powered collaborative search engine.
- Deep learning search platform with user contributions and blockchain incentives (Solana token).
- Full MLOps pipeline: ingestion, training, embedding updates, deployment.

Arm — ML Engineer Intern (Mar 2020 – May 2020, Cambridge, UK)
- R&D on generative models for speech synthesis.
- Python library for benchmarking ML pipelines on ARM processors.

CNRS, Centre for Brain & Cognition Research — Research Intern (Apr 2019 – Sep 2019, Toulouse, France)
- Spiking Neural Networks (SNNs) for audio analysis; RNNs with surrogate gradients in PyTorch.

SIDE PROJECTS

SLMs — https://github.com/romainzimmer/slm
Looped decoder-only transformer on TinyStories (RoPE, RMSNorm, SwiGLU; Muon + AdamW). Docker for Jetson Orin Nano.

Sudoku solver — https://github.com/romainzimmer/discrete-reasoning
Looped MLP-Mixer sudoku solver with commit loop, inner mixer iterations, learned halt head; trained on sudoku-extreme.

Teannamon — https://teannamon.com/
Tea brand with personalized AI assistant for recommendations and personalized tea creation.

EDUCATION
- École normale supérieure Paris-Saclay, M.S. Mathematics (2018 – 2019)
- Télécom Paris, M.S. Computer Science (2016 – 2018)

LINKS
LinkedIn: https://linkedin.com/in/romainzimmer
GitHub: https://github.com/romainzimmer
X: https://x.com/romzimmer
Correl8.ai: https://correl8.ai