Sohum Patnaik

Machine Learning Engineer

Sohum Patnaik is a Machine Learning Engineer at FSI, where he develops open-source tools and computational models to accelerate the transition to sustainable proteins.

Close-up of a smiling man with dark hair, beard, wearing a light blue button-up shirt, against a plain white background.
Every formulation is a decision. The right data and methods turn trial and error into design.
— Sohum Patnaik

Biography

Sohum Patnaik is a Machine Learning Engineer at Food System Innovations, where he builds open-source tools and computational models that help researchers and product developers design better sustainable protein products. His work focuses on modeling sensory attributes and optimizing formulation design—using data and machine learning to help make sustainable protein products tastier and more competitive in the marketplace.

At FSI, Sohum works at the intersection of artificial intelligence and food science, developing models that help predict how ingredients, processing techniques, and formulations influence taste, texture, and overall consumer experience. His research contributes to tools that enable faster experimentation and smarter product development across the sustainable protein ecosystem.

Prior to joining FSI, Sohum worked as a Data Scientist at Uber, where he used statistical analysis to measure user preferences and developed algorithms that optimized trade-offs within the Uber Eats marketplace. Earlier in his career, he worked at Booz Allen Hamilton, advising the National Park Service on data-driven investment strategies for infrastructure assets, helping balance competing priorities to better serve the agency’s mission.

Sohum holds a B.A. in Statistics from Williams College and is based in Seattle.

Areas of Expertise

  • Machine learning for food innovation

  • Sensory modeling & consumer preference prediction

  • AI-assisted formulation design

  • Data science for product optimization

  • Open-source research tools