Principal Research Engineer, Model Training & Post-Training

About the Company

Inflection AI is a Public Benefit Corporation empowering people with human-centered, emotionally intelligent AI. Inflection AI created Pi, the world’s first emotionally intelligent AI, to help people work through decisions, emotions, and challenges.

Responsibilities

• Own the model-improvement roadmap across capability, reliability, emotional intelligence, tool use, safety, latency, cost, and enterprise readiness
• Lead training and post-training strategy, including SFT, RLHF, DPO, GRPO, RLAIF, and reward modeling
• Drive model architecture and optimization decisions across transformer-based and hybrid architectures
• Lead large-scale training efforts on distributed GPU clusters (1,000+ GPUs)
• Define and execute data strategy across curation, mixture design, and synthetic data
• Build and improve evaluation and release-quality systems, including quality gates and regression detection
• Partner with infrastructure teams to improve distributed training reliability and observability
• Debug model behavior across the full stack from data to production feedback

Requirements

• Experience leading large-scale LLM, multimodal, or foundation-model training programs
• Deep experience with transformer-based models and distributed training systems
• Practical experience with post-training and alignment methods (SFT, RLHF, DPO, etc.)
• Experience operating or partnering on large-scale GPU clusters (1,000+ GPUs)
• Strong systems instincts regarding throughput, reliability, and fault tolerance
• PhD in Computer Science, Machine Learning, AI, or equivalent practical experience

Benefits

• Robust medical, dental, and vision options
• 401k matching program
• Flexible Time Off, 10 paid holidays, and 5 days sick leave
• Parental, Medical, and Family care leave
• Generous cell-phone, wellness, and office set up stipends
• Support of country-specific visa needs for international employees in the Bay Area