Summary

Research engineer role on h's inference team, building and operating the systems that serve multimodal agentic models in production. You will work across the inference engine and model serving layers to improve latency, throughput, and cost, and evaluate inference and hardware platforms. The role is hybrid, with an expectation to be in the office about 3 days per week.

Responsibilities

• Build and operate the inference stack
• Improve latency, throughput, and serving cost
• Research and implement inference techniques
• Co-design training-time decisions with models team
• Integrate inference into agentic ai products
• Evaluate inference, serving, and hardware platforms
• Communicate findings to stakeholders