AI Engineer

How you'll contribute

• Build and maintain the agent harness — the core runtime that orchestrates LLM calls, tool use, context construction, memory and guardrails — and make it reliable in production.
• Own the evaluation stack: design eval pipelines, golden datasets, LLM-as-judge and human-eval workflows; track agent performance, regressions and failure modes, and turn production failures into systematic improvements.
• Do prompt and behaviour engineering and retrieval/RAG at scale, defining what “good” looks like for agents completing complex accounting workflows end to end.
• Host and improve our own small models where it creates leverage: serving and inference of open‑source LLMs, help fine‑tuning and post‑training (SFT / DPO / RL), and document‑extraction models — for performance, cost and sovereignty.
• Work closely with Product teams and domain experts (accountants) so that fulfilling real user needs stays at the center of what we build.

What you can expect

Within one month:

• You'll learn everything about our company, our teams and our vision during the first onboarding week.
• You'll get familiar with our stack and AI tooling, and deliver a few small contributions that give you a concrete taste of our tools and processes.
• You'll be given time to meet your future stakeholders and gain a deep knowledge of our product and operations.

Within three months:

• You'll be in charge of items in our roadmap, defining and prioritizing your tasks autonomously.
• You'll be comfortable with our technical stack (Python, agentic frameworks, evaluation tooling, model serving and AWS).
• You'll contribute to larger cross‑team projects.

Within six months:

• You'll proactively contribute to the team's roadmap.
• You'll work with engineers and data practitioners on improving our stack, AI platform and evaluation practices.
• You'll share your learnings and best practices within the team.

And beyond, the team will keep growing with the company, which means:

• Opportunities to recruit and mentor new team members.
• Increased accountability in project leadership.
• Responsibilities to design and implement new processes, tools and best practices so your team works even more efficiently.

You're the right candidate if you

• Have 5–8 years of experience and are very strong in Python, with hands‑on experience building LLM and agentic systems at scale in production — prompting, tool use, context construction, RAG, and handling failure, state and reliability (not just calling a model API).
• Treat evaluation as a first‑class discipline: golden datasets, LLM-as-judge, human eval, A/B testing, and measuring agent quality, regressions and edge cases.
• Have a good grasp of applied LLM/ML and AI infrastructure (model serving, vector databases, cost and latency).
• Have exposure to model fine‑tuning and post‑training (SFT, DPO, RL).
• Have a balanced blend of technical, business and product skills, communicate well (including with non‑technical domain experts), and are fluent in English (French is not mandatory).

Who are we looking for?

• To speak English (levels are assessed and appreciated according to the department you're applying to).
• To be energized by an ever‑shifting work environment.
• To be highly collaborative (within your team and with other stakeholders).
• To be sufficiently experienced to prioritize business‑led actions in your day‑to‑day activity.

Benefits

• You will get 25 vacation days paid by Pennylane.
• You will have a competitive compensation package.
• You will get company shares.
• You will have a budget to turn your home into a more comfortable workspace, as well as a monthly allowance to work from a coworking space whenever you feel like it.
• Through our partner Gymlib, you will have access to fitness spaces in Europe and more than 300 wellness activities.
• You will have access to Busuu to perfect your English or your French.
• You will get the latest Apple equipment.
• Depending on the team and the requirements of the position, you can work remotely from your country of residence, as long as it is in Europe and within a maximum time difference of two hours from the CET time zone.
• We regularly come together for company events such as Tech Days and our annual company seminar.

If you are based in France, you'll have a French contract following French regulation, on top of the additional perks: 6 to 12 RTT, 5 weeks of PTO, lunch credits, Alan Blue healthcare cover, and regular events in cities where Pennylanes are mostly present (Lyon, Bordeaux, Nantes).

We're working on providing those last advantages to people based outside of France as well, though it can be more complex depending on the country.

Legal and Diversity Statements

We fully embrace diversity, equity and inclusion, and we're committed to providing equal employment opportunity regardless of gender, sexual orientation, origin, disabilities, or any other traits that make you who you are.

France Personal Data

Pennylane traite vos données pour gérer votre candidature et évaluer votre adéquation au poste. Si votre candidature n’aboutit pas, vos données peuvent être conservées 2 ans à compter de notre dernier échange ou de la clôture du recrutement afin de constituer & gérer un vivier de candidats. Vous pouvez vous opposer à tout moment et demander la suppression de vos données en écrivant à : Notre data policy

Personal Data

Pennylane processes your data to manage your application and assess your suitability for the position. If your application is unsuccessful, your data may be retained for 2 years from our last exchange or the closing of the recruitment process, in order to build and manage a candidate pool. You may object at any time and request the deletion of your data by writing to Learn more

Personenbezogene Daten

Pennylane verarbeitet Ihre Daten, um Ihre Bewerbung zu bearbeiten und Ihre Eignung für die Stelle zu beurteilen. Sollte Ihre Bewerbung nicht erfolgreich sein, können Ihre Daten bis zu 2 Jahre ab unserem letzten Austausch oder dem Abschluss des Rekrutierungsverfahrens gespeichert werden, um einen Kandidatenpool aufzubauen und zu verwalten. Sie können jederzeit Widerspruch einlegen und die Löschung Ihrer Daten beantragen, indem Sie an schreiben. Mehr erfahren