Machine Learning Research Engineer

About the role

Seqera is building a research engine on top of the Seqera Platform to accelerate scientific discovery. As a Machine Learning Research Engineer you will own the data and evaluation loop that continuously improves the engine, turning real‑world run data into high‑quality training corpora and post‑training small, task‑specific models.

Key responsibilities

• Design and implement an evaluation suite that defines performance metrics for an autonomous research agent.
• Build pipelines that transform messy deployment run data into training‑ready datasets.
• Post‑train open‑weight models (SFT, preference tuning, adapters) to outperform frontier API baselines while reducing cost and latency.
• Run repeated training cycles on evolving data and contribute to journal clubs to stay at the frontier of post‑training methods.
• Collaborate with the Founding Scientist to shape the roadmap, deciding which data unlocks which capabilities.

Required profile

• Experience with post‑training open‑weight language models, including SFT and at least one preference‑tuning method, deployed to real users.
• Hands‑on experience retraining models on changing data, handling regression and forgetting with techniques such as replay, data mixing, adapters, and evaluation gating.
• Proven ability to build custom evaluation harnesses, define metrics, and defend results.
• Fluent in Python and comfortable with an incomplete data or tooling environment.
• Willingness to work onsite in Barcelona or relocate there.

Required skills

• Python
• PyTorch
• Transformers
• TRL (Training Reinforcement Learning)
• PEFT (Parameter‑Efficient Fine‑Tuning)
• vLLM or equivalent inference engines
• Nextflow or workflow orchestration tools

What we offer

• Flexible working hours and an international environment with 25+ nationalities.
• Equity, private health and life insurance, and a home‑office allowance over $1,000.
• 23 vacation days plus additional holidays, plus a learning & development budget of $1,000 per year.
• Team retreats, bonding activities, and a culture that values opinions and impact.