Member of Engineering (Multimodality - Research Lead)
Remote • Remote (EMEA/East Coast) • FullTime
Posted 1mo ago
Remote Work Policy
Fully remote
Employment Type
FullTime
Categories
AI Research Engineer
About the job
Poolside is building a world where AI drives economically valuable work and scientific progress, aiming to accelerate the development of Artificial General Intelligence (AGI). We focus on reshaping the developer experience with agentic systems, coding assistants, and frontier models. This role is an opportunity to build and lead a new team focused on multimodality, specifically teaching our frontier coding model to understand and process image inputs. You will shape this critical capability from its early stages, leveraging our extensive resources including a powerful model factory, thousands of GPUs, and a strong research team. The initial focus will be on image input capabilities, such as interpreting designs, generating and verifying code, and understanding diagrams, with a long-term vision of evolving towards native multimodal understanding.
Responsibilities
- Own and define Poolside's multimodality strategy and capability adoption.
- Deliver the company's first image-input capabilities.
- Guide the transition from adapter-based approaches to native multimodality.
- Collaborate on custom evaluations and datasets for multimodal software engineering capabilities.
- Manage end-to-end experiments, from hypothesis to training at scale and analysis.
- Partner with various teams (evals, architecture, data, post-training) to integrate multimodality into Poolside models.
Requirements
- Experience in end-to-end training of production-grade Vision-Language Models (VLMs).
- Strong fundamentals in Large Language Model (LLM) training, including transformers and distributed training at scale.
- Proficient Python programming skills.
- Demonstrated pragmatism and high ownership, with a preference for the simplest effective solutions and comfort in ambiguous, greenfield environments.
- Leadership experience, particularly in building new initiatives from scratch (0-1) with broad ownership and hands-on contributions.
- Distinguished research background in VLMs or multimodal models.
- Understanding of the challenges associated with native multimodality.
- Experience building agentic computer use systems.
Benefits
- Fully remote work and flexible hours.
- 37 days per year of vacation and holidays.
- Health insurance allowance for employee and dependents.
- 16 weeks of flexible, full-pay parental leave.
- Allowances for well-being, continuous learning, and home office setup.
- Company-provided equipment.
- Frequent team get-togethers.
- Diverse and inclusive people-first culture.