Top AI Research Jobs in San Francisco Bay Area, CA

Research and improve voice AI models by identifying production weaknesses, defining evaluation metrics, curating data, fine-tuning deep learning systems, and rapidly shipping solutions. The role focuses on speech-to-text, text-to-speech, voice agents, model evaluation, customer-driven research, and production machine learning systems.

Build and deploy AI capabilities for Hex’s Notebook Agent and context engine. Responsibilities include running experiments, fine-tuning models, developing AI infrastructure and experimentation tooling, architecting AI-enabled systems, and shipping production experiences for data scientists and analysts. The role partners closely with product teams, spans greenfield initiatives, and offers opportunities for technical and team leadership.

The Director of Developer Relations leads the strategy for engaging developers around CoreWeave's AI SaaS offerings, creating content, fostering community, and improving the onboarding experience.

Lead and conduct ML research aligned to Sprinter Health's strategy: define research agenda, design rigorous experiments, develop novel methods and evaluations, publish and patent findings, translate research into production-ready tools, and collaborate with clinicians and cross-functional teams to validate and deploy clinically robust AI solutions.

Designs and prototypes scientific evaluations and reinforcement learning environments for frontier language models. Builds scalable pipelines that transform scientific assets into validated training environments, develops reward shaping and data collection workflows, and integrates experiments into model training infrastructure. Requires applied ML or computational experience in biology, chemistry, or materials science, strong Python skills, data pipeline expertise, research creativity, and software engineering fundamentals. Experience with LLM training, distributed computing, inference systems, and workflow orchestration is advantageous.

Designs reinforcement learning environments, multi-agent harnesses, and synthetic environment-generation pipelines for AI research. Responsibilities include prototyping tool interfaces, planning mechanisms, agent swarms, reward shaping, data collection, evaluation scenarios, and integration with training infrastructure. The role requires strong Python coding, reinforcement learning knowledge, research creativity, and software engineering fundamentals. Preferred experience includes scientific domains, developer-facing frameworks, distributed compute, LLM inference, and workflow orchestration.

Lead the AI x Lab Integration research agenda, developing models that understand scientific protocols, plan and execute experiments, diagnose failures, and adapt based on results. Recruit and mentor research scientists and engineers, establish rigorous evaluations, collaborate with experimental scientists and laboratory teams, and guide research in post-training, reinforcement learning, tool use, and learning from execution feedback.

Lead machine learning research from hypotheses through production-grade models. Train and refine large neural networks, develop reinforcement learning and synthetic-data methods, design rigorous evaluations, and operate distributed training systems. Collaborate with domain specialists and engineers to improve models, investigate failures, establish research priorities, mentor colleagues, and share results through papers or open-source work.

Develop agentic AI capabilities across Lumafield’s stack, including benchmarking LLMs, VLMs, and perception encoders; fine-tuning and post-training models; building deployed agents and evaluation harnesses; and creating evaluations that predict real-world performance. The role partners with product and engineering teams to improve agentic workflows and applies machine learning to specialized manufacturing, design, spatial, and perception data.

Leads Box’s applied AI research agenda by identifying high-impact opportunities, defining research priorities, partnering with engineering and AI labs on experiments, evaluating models and agents, translating findings into product improvements, and communicating evidence-based insights to customers, executives, media, and external audiences.

Lead research on action-conditioned world models and reward models for robot policy evaluation. Train generative and policy models, design rigorous evaluations, validate rewards against human judgments and task outcomes, and ship research into production systems. Collaborate directly with customers deploying robots, while shaping research direction across world models, robot learning, and real-world robotics.

Conduct foundational research on reinforcement learning and alignment for large language models. Develop preference datasets, reward models, verifiable rewards, and RL training pipelines using methods such as GRPO, RLHF, RLAIF, and DPO. Prototype scalable training recipes, collaborate with research and engineering teams, translate findings into customer-ready data products, and contribute to publications and internal knowledge sharing.