Machine Learning Engineer — Embedded AI at Pennylane

Are you looking to have an impact on the daily life of millions of entrepreneurs in France (and tomorrow in Europe)? Are you looking for a work environment that values trust, proactivity, and autonomy? If our Engineering principles align with your vision, Pennylane is the right place for you!

WHY this position is of utmost importance to reach our mission

Pennylane is becoming the financial OS of European SMEs and accounting firms. A growing share of the work done inside our product will be automated or assisted by AI. Our strategy operates at three layers: verticalized AI embedded in the product, an agent interface that makes Pennylane readable and actionable by AIs, and the Pennylane orchestrator agent itself.

The Embedded AI team owns the first layer. We build the specialized machine learning systems behind Copilot and Autopilot: invoice parsing, document classification, accounting suggestions, matching, scoring, Bookkeeping Autopilot, and Revision Autopilot.

Models are commoditizing fast. What makes the difference is our ability to frame the right problem, use Pennylane’s data and accounting knowledge, evaluate quality on real workflows, and operate reliable systems in production.

By joining us as a Machine Learning Engineer — Embedded AI, you will turn difficult accounting problems into trusted ML products used every day. You will combine strong ML engineering, product thinking, and production ownership to improve both automation and user trust.

HOW you will contribute as a Machine Learning Engineer — Embedded AI

You will join the Embedded AI team within our ML & AI organization. The team works closely with product squads and accounting experts, from initial exploration to production, monitoring, and continuous improvement.

• You will design and ship ML systems for document understanding, extraction, classification, matching, ranking, scoring, and recommendations.
• You will contribute directly to our Copilot and Autopilot experiences, including Bookkeeping Autopilot and Revision Autopilot.
• You will own the full lifecycle of your solutions: problem framing, data and labeling strategy, baselines, training, evaluation, deployment, experimentation, monitoring, and maintenance.
• You will turn user corrections and production failures into better datasets, models, and product behavior.
• You will define quality metrics that reflect real user value: precision and recall, automation coverage, straight-through processing, human correction rate, latency, and cost.
• You will partner with Product, Engineering, and accounting experts to understand workflows, define what “correct” means, and integrate ML naturally into the user experience.
• You will choose the simplest reliable approach for each problem — deterministic logic, classical ML, deep learning, or generative AI — rather than starting from a preferred model.
• You will help improve our shared ML engineering practices: reusable components, experimentation, observability, data quality, and reliable training and inference pipelines.
• You will stay ahead of the curve by monitoring emerging ML and AI techniques — including multimodal and generative models — and applying them when they create measurable value.

WHAT you can expect from your life at Pennylane

Within one month:

• You will learn about Pennylane, our users, our accounting workflows, and our AI vision during onboarding.
• You will get familiar with our ML stack, production systems, datasets, metrics, and ways of working.
• You will meet your product and engineering partners and contribute to a first scoped improvement.

Within 3 months:

• You will own an Embedded AI use case end-to-end, with clear offline and production metrics.
• You will have shipped a meaningful improvement to a Copilot or Autopilot capability.
• You will be comfortable with our technical stack, including Python, PyTorch, PySpark, Redshift, Airflow, AWS SageMaker, and our monitoring tools.
• You will use real user feedback and error analysis to prioritize the next iterations.

Within 6 months:

• You will lead larger cross-team ML projects and help shape the Embedded AI roadmap.
• You will improve the reliability, automation coverage, and maintainability of one or more production systems.
• You will share best practices and raise the bar for ML engineering, evaluation, and production ownership across the team.

And beyond: the ML & AI teams will continue growing with the company

This means opportunities to mentor new team members, lead major product initiatives, and help define the technical standards that make trusted AI possible at Pennylane.

Who are we looking for?

You may be a great fit if you:

• Have 5–8 years of experience and are very strong in Python.
• Have hands-on experience building and operating machine learning systems at scale in production — not only training models or running notebooks.
• Know how to frame an ambiguous product problem, establish a baseline, and select useful metrics before optimizing a model.
• Have strong experience in several of these areas: document AI, NLP, classification, ranking, recommendation, anomaly detection, deep learning, or generative AI.
• Care about data quality, observability, failure modes, cost, and long-term maintainability as much as model performance.
• Have a balanced blend of technical, business, and product skills, and communicate well with software engineers, product managers, and non-technical domain experts.
• Are fluent in English; French is not mandatory.

Nice to have:

• Experience with accounting, fintech, or other high-trust business workflows.
• Experience with human-in-the-loop systems, active learning, or learning from user corrections.
• Experience running ML systems at scale with strict latency, reliability, or cost constraints.
• Familiarity with multimodal or generative models for document understanding.