Technical Lead - Multimodal Transformers & AI Fusion

Kodiak AI; Mountain View (CA); On-site; USD 230,000 - 300,000; Full time

Benefits offered by this job

Equity

Bonuses

Medical coverage

Flexible PTO

Mountain View office

Dog-friendly office

Job summary

Kodiak Robotics, Inc. is seeking a senior technical leader to own the architecture direction of multimodal fusion that combines camera, LiDAR, and radar into unified representations, and to drive the project end to end from framing to production on trucks.

You will mentor engineers, set the pretraining strategy, and collaborate with perception, planning, and infra teams to meet latency and compute budgets while shaping scalable training pipelines.

Qualifications

• PhD with 5+ years of industry AI experience.
• MS/BS with 8+ years in AI or CS.
• Track record leading multi-engineer technical efforts.
• Expert in transformer architectures for multimodal data.
• Experience mentoring engineers.
• Proficiency in Python and PyTorch with distributed training.

Responsibilities

• Own the architecture roadmap for multimodal transformers.
• Lead end-to-end projects from framing to production.
• Drive research on cross-modal attention and token fusion.
• Set design standards for scalable training pipelines.
• Mentor engineers and raise research execution levels.
• Align model design with system latency and compute budgets.

Skills

Transformer architectures

Multimodal AI

Cross-attention

Token fusion

Distributed training

Mentoring

Python

PyTorch

System-level latency

Education

PhD

MS/BS with 8+ years AI/CS

Tools

Python

PyTorch

Job description

Kodiak Robotics, Inc. is seeking a senior technical leader to own the architecture direction of multimodal fusion that combines camera, LiDAR, and radar into unified representations, and to drive the project end to end from framing to production on trucks.

You will mentor engineers, set the pretraining strategy, and collaborate with perception, planning, and infra teams to meet latency and compute budgets while shaping scalable training pipelines.