Member of Technical Staff, ML Research Scientist
Omnifold trains custom AI models for each customer's supply chain - purpose-built systems that forecast demand, optimize decisions, and adapt continuously to a changing world. The research team is responsible for the core intelligence that makes this possible: developing new model architectures, curating proprietary data assets, and pushing the boundaries of what ML can do.
What makes this job interesting:
You will work on problems that frontier models can't solve. Supply chain dynamics require modeling physical systems and processes.
You will own the full research cycle, from hypothesis to production model, with direct visibility into real-world impact.
You will work at the intersection of machine learning models, optimization, LLM reasoning capabilities, and proprietary data - a combination few research teams are building
What you'll own:
Designing and training models for forecasting and optimization across complex, multi-variable supply chain environments
Building and curating proprietary data assets that carry signal about real-world physical and commercial systems
Integrating LLM knowledge and reasoning capabilities into purpose-built models to maximize accuracy and adaptability
Continuously improving model performance as market conditions shift (consumer sentiment, product launches, geopolitical changes, competitive dynamics)
What we're looking for:
Strong foundations in machine learning — experience developing and evaluating forecasting models, and LLM pipelines
Deep understanding of time-series forecasting, optimization, or related domains
Experience working with messy, heterogeneous real-world data
PhD or equivalent research experience preferred, but exceptional engineers with relevant industry experience will be considered
Comfort operating in a fast-moving, early-stage environment where research directly feeds production systems
Location: San Francisco (in-person, 5 days per week)
Omnifold’s Mission
Every bad forecast has a physical consequence. Unnecessary goods are manufactured, shipped, and stored. Emergency air freight is needed for misallocated products. Poor production planning means workers show up with nothing to do, or work frantic overtime. Inefficiency is everywhere.
Our mission is to eliminate waste and accelerate growth for every company with physical products.
Skills Required
- Strong foundations in machine learning
- Experience developing and evaluating forecasting models
- Experience developing LLM pipelines
- Deep understanding of time-series forecasting, optimization, or related domains
- Experience working with messy, heterogeneous real-world data
- PhD or equivalent research experience
- Exceptional engineers with relevant industry experience may be considered
- Ability to work in a fast-moving, early-stage environment where research feeds production systems
What We Do
Omnifold develops AI-powered planning software for complex supply chains and commercial operations. Its platform combines internal, external, and customer-specific data to model procurement, manufacturing, distribution, marketing, and sales dynamics, then uses prediction, optimization, language, and reasoning to improve forecasting. The system adapts to market changes and discovers growth strategies, helping businesses improve margins, cash flow, and operational decisions across complex enterprises.






