Engineering Lead – Multimodality, Research

This is a fully remote position, open to applicants in Europe.

📋 Description

• Take charge of Poolside's multimodal strategy and the adoption of its capabilities.

• Deliver our inaugural image-input functionalities.

• Plan and lead the transition from adapter-based systems to native multimodal solutions.

• Collaborate on specialized evaluations and datasets tailored for multimodal software engineering capabilities.

• Conduct experiments from start to finish: forming hypotheses, implementing solutions, scaling training, and analyzing results.

• Collaborate with evaluation, architecture, data, and post-training teams to integrate multimodality into Poolside models.

⛳️ Requirements

• Proven experience in the end-to-end training of production-grade vision-language models (VLMs).

• Solid foundation in large language model (LLM) training principles, including transformers and large-scale distributed training.

• Proficient in Python programming.

• Practical and ownership-driven: you seek the most efficient solutions and excel in ambiguous, greenfield environments.

• *Nice to have:*

• - Leadership and 0-1 experience with significant breadth, ownership, and hands-on contributions.

• - Notable research background in VLMs or multimodal models.

• - Awareness of the challenges associated with native multimodality.

• - Experience in developing agentic computer-use systems.

🏝️ Benefits

• Fully remote work with flexible hours.

• 37 days of vacation and holidays per year.

• Health insurance allowance for you and your dependents.

• 16 weeks of flexible, fully paid parental leave.

• Allowances for well-being, continuous learning, and home office setup.

• Company-provided equipment.

• Regular team gatherings.

• A diverse and inclusive, people-first culture.