Machine Learning Research Scientist

Etched is building the hardware for superintelligence.

GPUs and TPUs are flexible AI chips that can run many kinds of models: CNNs, RNNs, LSTMs, and more. But today, almost all AI workloads, from ChatGPT to self-driving cars, are done on one model architecture: transformers. Using flexible AI chips for transformers is very inefficient: <5% of the transistors on an H100 are used for matrix multiplication!

Etched is building a single-purpose chip exclusively for transformer inference. We only support transformers, but in exchange our chips have an order of magnitude more throughput and lower latency than an H100. With Etched, you can build products that would be impossible with GPUs, like tree-of-thought agents and ultra-low-latency audio chat bots.

Etched is looking for an ML Research Scientist to help our customers co-design models and highly specialized microchips. We believe that as the costs of the largest LLMs continue to climb, model-hardware codesign will become essential for keeping inference affordable.

Responsibilities:

• Design and implement deep learning architectures that will run efficiently on specialized silicon
• Understand new advances in NLP and how they will work with our chip architecture
• Accurately model performance of new transformer models on Etched’s specialized architecture
• Provide feedback to the architecture and hardware teams about advances in kernel parallelism strategies, both within one system and across multiple

Requirements:

• Experience designing architectures for large transformer models
• PhD in Computer Science, Electrical and Computer Engineering, Mathematics, or a related scientific discipline, or equivalent experience
• Deep and wide understanding of current research within large language models
• Deep understanding of how backpropagation is implemented and what alternatives might be.
• ​Ability to program with Python or another scripting language
• Passionate about AI scaling

Desired qualifications:

• Proficiency with GPU programming.
• Understanding of current techniques for efficient AI inference (including structured sparsity within memory, low precision floating point, and variants of attention).
• Experience with industrial-scale training runs (that cost >$10M)

Benefits:

• Competitive salary and equity package
• Full medical, dental, and vision packages, with 100% of premium covered
• Work with world-class people and state-of-the-art AIs everyday

Etched is committed to fair and equitable compensation practices. Compensation is determined based on your qualifications and experience. Compensation packages also include generous equity in Etched.

We are an equal opportunity employer and do not discriminate on the basis of race, religion, national origin, gender, sexual orientation, age, veteran status, disability or other legally protected statuses.