AI Architecture Researcher — State Space & Multimodal

Mat Vin; Greater London; Hybrid; GBP 90,000 - 130,000; Full time

Benefits offered by this job

Health Insurance

Commuter allowance

Flexible PTO

Meals & snacks

Visa sponsorship support

Job summary

Cartesia seeks a world-class ML architect to lead research on state space models, transformers, and related architectures. You will explore statefulness, conditioning, and memory to push real-time multimodal intelligence from cloud to device, collaborating across teams in a fast-growing startup environment.

The role requires a PhD or equivalent experience, a track record of impactful publications, and proficiency in PyTorch/TensorFlow.

Qualifications

• PhD in ML/AI or equivalent research experience.
• Strong publications in top ML/AI venues (e.g., NeurIPS/ICML/ICLR/CVPR).
• Experience researching or deploying advanced architectures (state space, transformers, RNN/CNN).
• Experience with multi-modal architectures and deployment constraints (cloud/edge).
• Proficient in Python and DL frameworks; ability to profile model performance.

Responsibilities

• Conduct groundbreaking research in neural network architecture design to advance state-of-the-art in alternative architectures.
• Design novel architectures to improve model quality, inference efficiency, and adaptability across deployment environments.
• Explore capabilities like statefulness, long-range memory, and conditioning mechanisms.
• Investigate how architectural decisions impact scalability, robustness, latency, and energy efficiency.
• Develop evaluation frameworks and tools for benchmarking research and production systems.
• Collaborate with cross-functional teams to translate research into scalable systems.

Skills

Architecture design

State space models

Transformers

RNN variants

CNN variants

On-device deployment

Multimodal architectures

Generative modeling

Statefulness & conditioning

PyTorch

TensorFlow

Model profiling

Education

PhD in ML/AI or related field

Tools

PyTorch

TensorFlow

Job description

Cartesia seeks a world-class ML architect to lead research on state space models, transformers, and related architectures. You will explore statefulness, conditioning, and memory to push real-time multimodal intelligence from cloud to device, collaborating across teams in a fast-growing startup environment.

The role requires a PhD or equivalent experience, a track record of impactful publications, and proficiency in PyTorch/TensorFlow.