Founding Member of Technical Staff - ML Infra

What You'll Do

• Build, maintain, and improve the algorithms and engineering systems used to post-train models for chip designs, focusing on improving training speed and reliability.
• Profile reinforcement learning and training pipelines to detect bottlenecks and implement optimizations for high-performance training setups.
• Collaborate closely with ML researchers to implement stable and fast versions of new finetuning recipes (like in RLHF/SFT) on different model architectures.

Qualifications & Skills

• PhD in Computer Science, EECS, Mathematics, or a closely related field, with specialization in Machine Learning, Systems, or Artificial Intelligence.
• Proven track record of building end-to-end ML pipelines, including data curation, preparation, and large-scale LLM finetuning (RLHF, SFT).
• Adept at diagnosing why training runs slow down, building instrumentation to monitor system health, and fixing complex issues in distributed environments.
• Results-oriented with a bias towards flexibility and impact, enjoying pair programming and picking up slack.
• Experience with implementing LLM finetuning algorithms (such as RLHF) and modifying systems based on model architectures.
• Foundation in Electrical/Computer Engineering and chip-design or verification processes (a plus).
• Publishations in top ML (NeurIPS, ICLR, ICML) or Systems (OSDI, SOSP) venues.
• Founding Engineer or early hire at an AI deeptech startup.