Architect, Multimodal Transformers for Autonomous Systems

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

Benefits offered by this job

Competitive compensation including equ

Annual bonuses

Excellent medical, dental, vision

Dog-friendly office

Free catered lunch

EV charging

Job summary

Kodiak Robotics, Inc. is seeking a senior technical leader to own the architecture roadmap for multimodal transformers fusing camera, LiDAR, and radar. You will lead end-to-end projects from framing to production and set architectural direction for scalable training pipelines.

With a PhD or MS/BS and 8+ years of experience, you will mentor engineers, drive cross-modal research, and ensure AI systems meet safety and latency requirements in real-world trucking applications.

Qualifications

• PhD with 5+ years of industry experience or MS/BS with 8+ years in AI/CS.
• Track record of leading multi-engineer technical efforts from research through production.
• Deep expertise in transformer architectures, particularly multimodal or multi-stream settings.
• Strong command of cross-attention, token fusion, and modality alignment techniques.
• Experience mentoring engineers and growing technical talent.
• Proficiency in Python and PyTorch, including large-scale distributed training and mixed-precision optimization.
• Ability to influence technical direction across teams without formal authority.
• Passion for building AI that reasons over the full breadth of sensory input to operate safely in the real world.

Responsibilities

• Own the architecture roadmap for multimodal transformers that fuse camera, LiDAR, and radar into unified representations.
• Lead the project end to end: problem framing, experiment design, implementation, and production deployment.
• Drive research direction on cross-modal attention, token fusion strategies, and efficient multi-stream tokenization; decide what gets built.
• Set the technical bar through design reviews, code reviews, and architectural decisions on scalable training pipelines.
• Mentor junior and mid-level engineers, and raise the level of research execution across the org.
• Define the pretraining strategy, including self-supervised and contrastive objectives for transferable multimodal representations.
• Partner with perception, planning, and infrastructure leads to align model design with system-level latency and compute budgets.

Skills

Transformer architectures

Multimodal AI

Python

PyTorch

Distributed training

Mentoring

Cross-attention

Token fusion

Leadership

Education

PhD

MS/BS + experience

Tools

PyTorch

Job description

Kodiak Robotics, Inc. is seeking a senior technical leader to own the architecture roadmap for multimodal transformers fusing camera, LiDAR, and radar. You will lead end-to-end projects from framing to production and set architectural direction for scalable training pipelines.

With a PhD or MS/BS and 8+ years of experience, you will mentor engineers, drive cross-modal research, and ensure AI systems meet safety and latency requirements in real-world trucking applications.