Qualifications

• 5-7 years of hands-on experience with production-level foundation models and multimodal models.
• Strong intuition for understanding model behavior, evaluation methodologies, and identifying failure modes.
• Proven skills in fine-tuning and training pipelines, along with dataset construction expertise.
• Familiarity with reinforcement learning from human feedback (RLHF), synthetic data generation, or active learning techniques is a plus.
• Proficiency in Python-based ML stacks, and experience with Typescript-based production systems.
• Action-oriented mindset with a focus on rapid experimentation, measurement, and iteration.
• Ownership mentality with a commitment to achieving real-world outcomes.

Responsibilities

• Train, fine-tune, and deploy advanced multimodal models for practical applications.
• Design and implement frameworks for rigorous model performance evaluation and subsequent iteration.
• Build and scale data flywheels to collect and generate high-quality datasets for continuous model improvement.
• Develop systems that support live learning and adaptive feedback incorporation in production.
• Apply techniques like negative mining to enhance model resilience against edge cases.
• Manage the full lifecycle of model development from experimentation to monitoring in production.
• Collaborate closely with engineering and product teams to integrate models into reliable, high-performance systems.

Benefits

• Opportunity to join a pioneering team focused on building an enduring company.
• Directly contribute to food accessibility for millions of households.
• Work with state-of-the-art AI technology using modern frameworks.
• Competitive salary with substantial equity potential.
• Comprehensive health benefits including medical, dental, and vision coverage.
• In-person collaboration alongside an experienced founding team in San Francisco.