About Figma
Figma is a collaborative interface design tool that runs in the browser, enabling teams to design, prototype, and gather feedback in real time.
About the role
People store millions of files, components and community templates in our product, and finding the right one is harder than it should be. Search results today rely mostly on text matching, which fails when someone types “blue checkout button” and the component is called btn/primary/lg.
We are building a new ranking system that understands design content, not just file names. As a Machine Learning Engineer in Amsterdam, you will join the Search & Discovery team, a group of six engineers and one researcher, and help take this system from prototype to production.
What you’ll do
- Train and evaluate ranking models that combine text, visual features and usage signals.
- Build the offline evaluation pipeline, including:
- labelled query sets for at least 12 common search intents
- metrics such as NDCG and recall at 10
- regression checks that run before every model release
- Ship models to production with latency below 80 ms at the 95th percentile.
- Design online experiments with the product team and analyse the results.
- Write clear model documentation, covering training data, known limitations and privacy considerations.
What we’re looking for
- Four or more years building machine learning systems that run in production.
- Strong Python and solid experience with a deep learning framework.
- Practical knowledge of learning-to-rank, embeddings or recommendation systems.
- Good habits around data: you check for leakage, you version datasets, and you can explain why a model improved.
- Comfortable working with privacy constraints. Customer files are never used for training without explicit consent.
Nice to have
- Experience with multimodal models that combine images and text.
- Familiarity with vector databases and approximate nearest neighbour search.
- Published research or open source work in information retrieval.
Compensation and benefits
- Base salary of €95,000–€120,000, plus equity.
- 30% ruling support for eligible international hires, and relocation assistance up to €8,000.
- 27 days of paid holiday plus Dutch public holidays.
- An 8% holiday allowance, paid every May.
- A €1,500/year learning budget and access to GPU resources for side experiments.
- Hybrid: our office near Amsterdam Centraal is open 08:00–19:00, and the team meets in person on Mondays and Thursdays.
How to apply
Send your CV and a short description of a model you shipped, including one metric that improved and one trade-off you accepted. Links to papers or repositories are welcome.
Interview steps:
- A call with the engineering manager (30 minutes).
- A technical interview on ranking and evaluation.
- A coding session in Python, focused on data processing rather than puzzles.
- A final conversation with the head of Search.
The deadline is 10 December 2026. You can read about the team’s culture on the company profile, or browse other AI and machine learning jobs.
