AI Systems Architect – Multimodal Models on Novel Compute

Unconventional, Inc.; Mountain View (CA); On-site; USD 180,000 - 320,000; Full time

Benefits offered by this job

Health benefits

401k matching

Unlimited PTO

Complimentary meals

Job summary

Unconventional, Inc. is seeking a Member of Technical Staff, AI Systems, Algorithms & Models to design and scale model architectures that merge novel computational primitives with conventional neural components for image generation, language, video, and world models on our unique compute substrates.

You will co-design next-generation AI models (diffusion, flow, energy-based models, Neural ODEs, Deep Equilibrium Models), test scaling laws, and develop generalized learning frameworks across

Qualifications

• MS/PhD in a quantitative field (AI/ML, CS, Physics, EE or Applied Math).
• Deep expertise in modern ML model architectures and empirical analysis.
• Proven ability to design, implement and scale models across tasks and modalities.
• Experience with PyTorch or JAX and custom training loops.

Responsibilities

• Co-design and evaluate next-generation AI models (diffusion, flow, energy-based models, Neural ODEs, Deep Equilibrium Models).
• Collaborate across teams to integrate core modeling components with both conventional and novel hardware.
• Develop scaling laws and methodologies (muP) to push model capability on novel substrates.
• Translate algorithmic requirements into concrete system/hardware specifications.

Skills

ML theory & training

Model scaling

Cross-disciplinary collaboration

Transformers

Education

MS/PhD or equivalent

Tools

PyTorch

JAX

Job description

Unconventional, Inc. is seeking a Member of Technical Staff, AI Systems, Algorithms & Models to design and scale model architectures that merge novel computational primitives with conventional neural components for image generation, language, video, and world models on our unique compute substrates.

You will co-design next-generation AI models (diffusion, flow, energy-based models, Neural ODEs, Deep Equilibrium Models), test scaling laws, and develop generalized learning frameworks across