Real-World AI Research Intern (PhD)

Location: Remote or Palo Alto, CA

Duration: 12–16 weeks (flexible)

Compensation: Paid, competitive

Start: Rolling

Palona builds real-world AI systems that operate continuously in production. Our work focuses on AI agents that perceive, reason, remember, and act in physical environments, starting with restaurants as a constrained but high-signal domain.

We are interested in research that survives contact with reality: partial observability, delayed effects, noisy signals, non-stationarity, and long-horizon outcomes.

This internship is for PhD students who want to work on applied research problems grounded in deployed systems.

You will work on questions that arise from live AI agents operating in the real world, where clean assumptions break and system behavior must be understood over time, not just measured offline.

We are looking for candidates with deep research experience in at least one primary area, and working familiarity with adjacent areas.

• Reinforcement learning, planning, or control
• POMDPs or decision-making under partial observability
• Credit assignment with delayed and sparse rewards
• Long-horizon optimization

Relevant signals:

• Publications in RL, planning, or control venues
• Experience implementing and evaluating decision-making agents

• Latent state models for dynamic environments
• Temporal abstraction and hierarchical representations
• Persistent memory or state tracking
• Modeling environments that evolve over time
• Research on state-space models, memory-augmented models, or temporal representations

• Belief state estimation
• Uncertainty modeling in dynamic systems with incomplete or noisy information
• Research in probabilistic modeling, causal inference, or dynamic systems

• Vision-language models
• Learning from asynchronous, noisy, or partially missing modalities
• Sensor fusion or multimodal representation learning
• Publications or projects involving multimodal models
• Experience working with real-world (not synthetic-only) data

Projects are scoped based on your expertise and may include:

• Designing world state representations that persist across time, entities, and events
• Modeling cause and effect in real operational workflows
• Building reasoning systems that operate with partial observability and delayed outcomes
• Developing evaluation methods for agents running in production
• Translating research ideas into systems that are deployed and iterated on

You will collaborate closely with senior researchers and engineers and see how your work affects system behavior in the real world.

• Strong problem formulation skills
• Ability to connect theory with implementation
• Comfort working with ambiguity and evolving research questions
• Thoughtful evaluation and reflection on system behavior over time

• Exposure to research problems shaped by real deployment constraints
• End-to-end ownership from research idea to production impact
• Close mentorship from experienced AI practitioners
• Opportunity for continued research collaboration beyond the internship

Please include:

• CV
• Google Scholar or publication list
• A short statement (1–2 paragraphs) describing:
• Your primary research focus
• Why you are interested in real-world, production-grounded AI research

Required

• Current PhD student in CS, AI, ML, Robotics, or a closely related field
• Strong research record (publications or equivalent contributions)
• Hands-on experience implementing research ideas in code
• Solid foundations in machine learning and statistical reasoning

Preferred

• Experience with deployed or real-world ML systems
• Prior industry or applied research experience
• Strong Python and ML systems skills