Foundational AI Research Intern -Deep Learning PhD ONLY

About Expert Intelligence

Expert Intelligence builds the decision layer for regulated analytical labs.  The company positions its platform around two products in one system: EI Flow, which automates in-batch decisions from instrument data to signed reports, and EI Signal, which reveals trends across instruments, methods, and sites.

Website: https://www.expertintelligence.ai/

Role Summary

Expert Intelligence is hiring an in person Foundational AI Research Intern / Research Resident for Summer 2026 in Santa Clara, California. This role is for exceptional final-stage PhD candidates, recently completed PhDs, and postdoctoral researchers who already operate at a high level in mathematically rigorous generative modeling. This is not a general deep learning internship. It is intended for candidates who can move fluently from equations to implementation in modern transport-based, geometry-aware, and constrained generative systems.

Areas You May Work On

• flow matching, conditional flow matching, and rectified flow
• unified generative modeling across diffusion, flow matching, consistency, and related transport-based methods
• constrained generative modeling under geometric, physical, or structural constraints
• discrete and continuous transport-based generation
• geometry-aware and manifold-aware generative learning
• equivariant generative modeling for structured scientific domains
• optimization methods for stable large-scale generative learning
• efficient inference, trajectory refinement, and fast-sampling generative systems
• scientific generative modeling for molecules, proteins, sequences, inverse problems, or other structured domains
• JAX-based implementation of mathematically grounded research systems

Mathematical Rigor Expected

This role requires exceptional mathematical maturity. Candidates should already be comfortable deriving and reasoning about transport-based generative objectives, continuity equations, continuous-time generative dynamics, optimal transport, constrained optimization, information geometry, natural gradients, equivariant parameterizations, and manifold-aware learning. This is not a general deep learning role. It is intended for candidates who can work fluently from equations to implementation in advanced generative modeling systems.

Please DO NOT apply, 

unless you are already comfortable with advanced generative modeling, mathematical derivation, optimization theory, information geometry, and implementing mathematically grounded research systems in JAX or similar high-performance research stacks.

Interview Standard

Our interview process is designed to measure raw technical depth. Candidates should expect to derive nontrivial objectives and updates from first principles, explain the mathematical structure of a generative method rather than describe it only at a high level, reason about optimization, stability, and geometry in model design, and translate equations into correct and numerically stable implementation. No AI tools will be allowed during technical evaluation.

In Person Requirement

This role is in person in Santa Clara, California. Please do not apply if you cannot work onsite in Silicon Valley.