AI Engineer, Training Data Infrastructure

Avala; San Francisco (CA); Hybrid; USD 150,000 - 230,000; Full time

Benefits offered by this job

Competitive salary

Equity

Health insurance

Dental and vision

FSA

Unlimited PTO

Job summary

Avala is building multimodal training data infrastructure in San Francisco to power Physical AI for autonomous vehicles and robotics. You will design and implement large-scale data pipelines that process petabytes of video, sensor, and action data from a global network of annotation experts.

The role emphasizes scaling models, distributed processing, and modern DL frameworks, offering competitive compensation, equity, and health benefits in a fast-moving AI hardware context.

Qualifications

• 3+ years experience scaling and optimizing large AI models.
• Proficiency in Python with strong software engineering practices.
• Deep learning fundamentals: optimization, loss functions, neural network architectures.
• Experience with PyTorch, TensorFlow, or JAX.
• Strong expertise in distributed computing and parallel processing.
• Experience with petabyte-scale data processing pipelines.

Responsibilities

• Create multimodal training datasets from diverse robotics and AV data sources.
• Develop infrastructure for training large multimodal models at scale.
• Design data processing pipelines for video, LiDAR, radar, and sensor fusion.
• Build downstream evaluation systems to guide model tuning and dataset composition.
• Run scaling experiments for model size, dataset quality, and training efficiency.

Skills

Python

Distributed computing

Deep learning

Model scaling

Petabyte-scale data

Tools

PyTorch

TensorFlow

JAX

Data pipelines

Job description

Posted 10 months ago San Francisco, CA full-time

About This Role

Build the multimodal training data infrastructure that powers Physical AI systems for autonomous vehicles and robotics. Design and implement large-scale data pipelines that process petabytes of video, sensor, and action data from our global network of annotation experts.

Key Responsibilities

• Create multimodal training datasets from diverse robotics and AV data sources
• Develop infrastructure for training large multimodal models at scale
• Design data processing pipelines for video, LiDAR, radar, and sensor fusion
• Build downstream evaluation systems to guide model tuning and dataset composition
• Run scaling experiments for model size, dataset quality, and training efficiency

Requirements

• 3+ years experience scaling and optimizing large AI models
• Proficiency in Python with strong software engineering practices
• Deep learning fundamentals: optimization, loss functions, neural network architectures
• Experience with PyTorch, TensorFlow, or JAX
• Strong expertise in distributed computing and parallel processing
• Experience with petabyte-scale data processing pipelines

What We Offer

Competitive salary and meaningful equity in the team building Fleet Learning Infrastructure for Physical AI

Full health, dental, and vision coverage, plus FSA and unlimited PTO

Work on the closed-loop data infrastructure that the next generation of autonomous systems learns from

A team that built Tesla Autopilot's data infrastructure, now building something bigger