Member of Research Staff, Post-Training

Modal • San Francisco, CA

1d • $150,000 - $350,000 • Onsite

About The Position

AI needs a new infrastructure layer, and Modal is building it. The company is defining the decade by rebuilding the infrastructure layer underneath AI workloads. Modal's customers include category-defining companies like Lovable, Ramp, Cognition, DoorDash, and Suno, relying on Modal for instant GPU access, sub-second container starts, and native storage for serving low-latency inference, fine-tuning models, and accessing production-ready sandboxes at scale. Modal recently raised a $355M Series C at a $4.65B valuation, led by General Catalyst and Redpoint Ventures, and has crossed $300M+ ARR, growing fivefold since September. The team comprises creators of popular open-source projects, academic researchers, international olympiad medalists, and experienced engineering and product leaders.

Modal is building a platform that covers the entire lifecycle of an LLM: training, deployment, and production observation. They already support multi-node training, elastic inference, sandboxes, and distributed volumes, with control over the underlying infrastructure. The company is seeking individuals with research depth in post-training to complement their systems and product development efforts.

Requirements

• A research-leaning background in post-training LLMs, with demonstrable work.
• Sufficient product sense to identify frontier techniques that are relevant to users versus those that remain academic.
• A track record of shipping research that is utilized by others, whether in an academic lab or industry.
• The drive to independently take a research bet from conception to completion, working collaboratively and transparently with the team.
• Ability to work in-person in either the NYC or San Francisco office.

Responsibilities

• Own end-to-end post-training research bets: async and agentic RL, on-policy distillation, long-context RL, small routing models, and other areas as defined by the research agenda.
• Work directly with customers alongside Forward Deployed Engineers to train models and integrate learnings back into research.
• Carry forward and expand collaborations with outside research labs, such as the work with ZLab on DFlash.
• Collaborate with engineering to translate frontier post-training techniques into products, including an opinionated post-training framework, distributed-training approaches (DiLoCo, evolutionary strategies), and online training for deployed models.
• Help shape the research agenda, with work guiding future directions.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed