Research Engineer / Scientist (LLM Modeling)

TL;DR

Research Engineer/Scientist (LLM Modeling): Advance the frontier of large-scale language model building and deployment with an accent on pre-training, post-training, RL, and specialized architectures for reasoning and decision-making in critical industries. Focus on prototyping and scaling training pipelines, experimenting with model architectures and optimization, and transitioning research into production features.

Location: New York City or Boston

Company

Transforming critical institutions like healthcare, manufacturing, and energy with applied AI through forward-deployed expertise, in-house Mosaic toolkit, and partnerships with Anthropic, McKinsey, AWS, and General Catalyst.

What you will do

• Identify impactful problems and design research strategies for pre-training, post-training, RL, and specialized model development.
• Prototype and scale training pipelines for large language models, experimenting with architectures, optimization, and post-training strategies.
• Contribute to infrastructure for high-performance distributed training.
• Conduct large-scale in-the-wild evaluations driving significant business value.
• Partner with applied AI engineers to integrate research into Mosaic platform features.
• Communicate research outcomes to technical and non-technical stakeholders.

Requirements

• MS/PhD in Computer Science, ML, related field, or equivalent experience.
• Depth in LLM and ML fundamentals from optimization to large-scale training and Reinforcement Learning.
• Comfortable implementing and debugging large-scale ML systems.
• Proven track record of execution and excellent communication with technical/non-technical stakeholders.
• Motivated by impact in critical industries: healthcare, supply chains, energy, finance.

Culture & Benefits

• Dream bigger: Tackle ambitious problems with optimism and responsibility.
• Heart in the game: Care deeply about impactful work without monitoring hours.
• Win for the customer: Focus on delivering outcomes as an engineer.
• Make the call: High-agency decisions with high-bandwidth communication.
• Intensity with kindness: Excellence, candor, prioritization, and urgency balanced with kindness.