What you'll do:

• Train, fine-tune, and serve open-source LLMs (Qwen, Mistral, etc.) that power real products
• Design experiments, evaluate results, and ship solutions that win in A/B tests
• Work on the full ML lifecycle: from training runs to production serving infrastructure

Preferred qualifications & traits:

• You know how to write good Python code from experience building real projects (or close to it). Bonus points if you've worked with distributed systems.
• You've actually used PyTorch and Huggingface and you also have an idea of what's going on under the hood.
• You understand ML fundamentals because you've trained models yourself and understand the core concepts.
• You can make sense of research by reading papers, getting the key ideas, and turning them into working code (with the help of your AI assistant).
• You think for yourself and can come up with ideas.
• You're genuinely curious and actually care about understanding how things work and get excited by new ideas
• You love learning new tech and applying it.
• You work well with others by communicating clearly, not being shy to ask questions, and helping your teammates.
• You're flexible when priorities shift and okay with switching from research to data processing when needed.
• You're friends with AI assistants and use them regularly for coding and writing.
• You can commit to 6-12 months working with us in Amsterdam (minimum 3 days in the office weekly). We'll work with your academic schedule.

What we offer:

• Learn from the best: Work with AI engineers who've published at NeurIPS, released top models on Huggingface, and built production systems at scale
• Ship from day one: Your code goes into production quickly, not after months of onboarding
• Real infrastructure: Access to H200 GPU clusters, large-scale datasets, and all major LLM APIs (OpenAI, Anthropic, Google)
• Unique data at scale: Access to billions of customer interactions from 500+ million users across our e-commerce platforms, with millions of products
• Grow your career: People who excel here tend to stick around
• Junior Engineer compensation: We pay competitively for the impact you'll have
• Best-in-class tools: MacBook Pro, access to all major LLM APIs and premium coding assistants
• Great workspace: Modern office in Amsterdam South with free barista coffee and NS travel card for your commute