Machine Learning Research Scientist

Palo Alto, California
Engineering/Research
Full Time
Fathom Computing is developing high-performance machine learning computers built to run both training and inference workflows for large-scale artificial neural networks (ANNs).  As computing is largely limited by data transfer Fathom’s approach combines the power of CMOS electronics with advantages of optical data movement for performance far beyond what electronics-only computers are capable of.

We’re seeking a talented Machine Learning Research Scientist with strong first-principles understanding of neural networks to collaborate with our optics and electronics teams in design and implementation of novel optoelectronic hardware.

Areas of contribution

    • Implementing novel machine learning algorithms on our unique hardware
    • Considering several layers of abstraction, all the way from lower level circuits through instruction set architecture
    • Designing new ML algorithms for future hardware systems
    • Developing, adapting, and mapping general machine learning algorithms based on features of our hardware
    • Inventing new models that combine unsupervised and supervised learning, with the kind of creativity usually reserved for blue-sky research projects

Requirements

    • BS/MS/PhD, or equivalent experience in CS, EE, or related fields (e.g. statistics, applied math, computational neuroscience)
    • Deep passion and fundamental understanding of design, algorithms, and data structures in modern machine learning and AI
    • Strong understanding of the fundamentals of neural networks and common general algorithms including RNN, CNN, RL
    • Strong analytical skills (probability, optimization, etc.)
    • Experience working with large models
    • Depth and breadth of knowledge in the field, including recent research results
    • Knowledge of current frameworks (e.g. TensorFlow, PyTorch, etc.)
    • Excellent communication skills and ability to collaborate on a complex cross-disciplinary system
    • Drive to build something that hasn't been built before

You'll do well here if...

    • You enjoy thoughtful discussions fueled by problem-solving and logic
    • You're comfortable both leading and contributing individually
    • You're excited about the future of ML hardware
    • You enjoy teaching and learning from an interdisciplinary team


We highly encourage submission of a cover letter, just tell us why you're here :)