AI Engineer
Sign up free to see how well your resume matches this role.
What you'll do
- Train, fine-tune, and evaluate our proprietary language models — including the small models that run on-device.
- Curate and maintain the datasets, golden sets, and evaluation harnesses those models are judged against.
- Ship model updates into production and measure the result, rather than stopping at a promising notebook.
- Own the health of the AI features in production: quality, cost, and latency — and catch silent regressions before users do.
- Debug AI incidents end to end, from retrieval quality to prompt drift to model behaviour on a client's machine.
- Keep our models, prompts, and pipelines current as the underlying tooling and APIs evolve.
- Partner with product managers and engineers on new AI use cases, and turn the promising ones into quick, honest prototypes.
- Bring the evidence — evals, cost estimates, latency numbers — that lets the team decide what's worth shipping and what isn't.
- Explain AI trade-offs clearly to non-technical colleagues so decisions get made quickly.
- Document what you build and why, so the practice compounds instead of living in one person's head.
- Coach the more junior engineers and interns who work with you on AI topics.
What they're looking for
- 3–5 years in ML / AI engineering, with models or AI features you shipped and then operated in production.
- Hands-on with LLM fine-tuning, RAG, and embeddings — and with evaluating them (golden sets, LLM-as-judge, regression tests).
- Experience monitoring and optimising AI systems in production for cost, latency, and quality.
- Solid software engineering fundamentals — tests, CI, observability — you operate what you ship.
- Comfortable in Python and able to work across a TypeScript product codebase.
- Execution-first: you would rather ship a measured improvement than pitch an ambitious redesign.
- Pragmatic and product-minded — user impact over technical novelty.
- Autonomous on your scope, and clear about when to escalate an architecture or cost decision.
- Excellent command of French and English, both technical and professional.
Nice to have
- Experience with small language models (SLMs) and on-device / on-prem model deployment is a strong plus.
Summarised by NextRaise from the employer’s description, which follows in full below.
Full description from employer
Location: Paris, Châtelet – Hybrid (2 days on-site / 3 days remote)
Contract: Full-time (permanent)
Team: Technology
Compensation: €56,000–€65,000 gross annual package (base + 8% target variable) + BSPCE (equity), depending on step
🚀 About Mendo — and why this matters
We're living through the biggest technological shift since the internet. Generative AI is changing how people work — yet most teams barely scratch the surface, unsure how to turn the hype into real, everyday value. Closing that gap is exactly what Mendo does.
Mendo is the AI Transformation Platform for enterprise clients. Rather than adding yet another tool to the stack, we work right inside the AI tools teams already use — Microsoft 365 Copilot, ChatGPT, Gemini, Claude, and internal agents. From there, we guide people toward proven use cases, certify their skills, and give organizations a clear view of what's actually driving value.
And it's working. We support 100+ enterprise clients — including EY, PwC, Crédit Agricole, SNCF, La Poste, ENGIE, Groupe SEB, and Novo Nordisk — with 200+ use cases live, a user NPS of 64, and 3x GenAI ROI within months. We've raised €12M following our Series A and we're scaling fast.
Join us and you won't just ship software — you'll shape how leading organizations adopt the defining technology of our era, and help keep people at the center of the AI revolution rather than left behind by it. We believe humans have a vital role to play in this revolution, and we're building the platform that keeps them there.
🧭 Core mission
Make Mendo's AI work, every day. We already ship AI across the product — proprietary small language models that run on people's own devices, retrieval and classification pipelines, and the analytics that tell organizations what's actually driving value. Your job is to execute: train and evaluate the models, keep the features we've shipped healthy in production, and be the hands-on AI partner the product squads turn to when they're shaping what comes next.
You'll work alongside our CTO and the engineering squads, own the AI workstream for a product area end to end, and become its AI referent. This is an individual-contributor role with no direct reports — we need someone autonomous who ships and operates, not someone who wants to run a team.
🛠️ Key responsibilities
1. Train and evaluate our models
Train, fine-tune, and evaluate our proprietary language models — including the small models that run on-device.
Curate and maintain the datasets, golden sets, and evaluation harnesses those models are judged against.
Ship model updates into production and measure the result, rather than stopping at a promising notebook.
2. Maintain what we've built
Own the health of the AI features in production: quality, cost, and latency — and catch silent regressions before users do.
Debug AI incidents end to end, from retrieval quality to prompt drift to model behaviour on a client's machine.
Keep our models, prompts, and pipelines current as the underlying tooling and APIs evolve.
3. Brainstorm and prototype with the squads
Partner with product managers and engineers on new AI use cases, and turn the promising ones into quick, honest prototypes.
Bring the evidence — evals, cost estimates, latency numbers — that lets the team decide what's worth shipping and what isn't.
Explain AI trade-offs clearly to non-technical colleagues so decisions get made quickly.
4. Raise the bar on your scope
Document what you build and why, so the practice compounds instead of living in one person's head.
Coach the more junior engineers and interns who work with you on AI topics.
🧱 Tech stack
Languages: Python for AI/ML work, TypeScript across the product.
On-device / SLMs: GGUF models running in-browser via WebLLM and wllama (llama.cpp compiled to WASM), packaged into our browser and desktop connectors.
LLMs & GenAI: OpenAI APIs, retrieval-augmented generation, and embeddings, integrated into Microsoft 365 Copilot, ChatGPT, Gemini, Claude, and internal agents.
ML tooling: sentence-transformers, HuggingFace Transformers, scikit-learn, and TensorFlow, served through a Python service.
Data: MongoDB and Redis.
Infra & deployment: Azure / AKS with Helm, Docker containers, GitHub Actions CI/CD, Datadog observability.
AI-augmented engineering: Claude Code / Cursor, MCP tooling, and internal AI agents are part of the daily workflow — not a bonus.
This stack is always evolving, and we're counting on you to help us evolve and improve it.
🧩 Required skills
Experience & expertise
3–5 years in ML / AI engineering, with models or AI features you shipped and then operated in production.
Hands-on with LLM fine-tuning, RAG, and embeddings — and with evaluating them (golden sets, LLM-as-judge, regression tests).
Experience monitoring and optimising AI systems in production for cost, latency, and quality.
Solid software engineering fundamentals — tests, CI, observability — you operate what you ship.
Comfortable in Python and able to work across a TypeScript product codebase.
Experience with small language models (SLMs) and on-device / on-prem model deployment is a strong plus.
Ways of working
Execution-first: you would rather ship a measured improvement than pitch an ambitious redesign.
Pragmatic and product-minded — user impact over technical novelty.
Autonomous on your scope, and clear about when to escalate an architecture or cost decision.
Excellent command of French and English, both technical and professional.
Mindset
Natural rigor and a feel for evaluation-driven development.
Genuine curiosity for a fast-moving AI landscape — without chasing hype.
Strong team spirit and a sense of priorities in a fast-moving scale-up.
Alignment with our values: Learning, Innovation, Excellence, Kindness, Transparency.
📈 What we offer
Competitive package: €56,000–€65,000 gross annual package (base + 8% target variable) depending on step, plus BSPCE (equity) under our Series A policy.
Real impact: the models you train and the features you maintain run for 100+ enterprise clients.
Real AI, not slides: proprietary on-device models, production RAG, and a codebase where AI is already shipped — you'll work on it, not argue for it.
A learning environment: a team passionate about new technologies and generative AI.
A stimulating workplace: a young, supportive culture where everyone helps each other.
Flexibility: a hybrid 2-day/3-day remote model.
Meaningful mission: helping make AI accessible and impactful for everyone.
Company
Company facts come from this company's own listings. We only show what the postings themselves carry.