Rider Simulation & LLM Agent Intern

CHEManager International; San Francisco (CA); Híbrido; USD 88.000 - 94.000; Tempo integral

Benefícios oferecidos por esta vaga

Great medical, dental, and vision

Mental health benefits

2 days paid time off + 3 days sick

401(k) plan

Subsidized commuter benefits

Lyft Pink - Ridership Program

Resumo da vaga

Lyft is seeking an Applied Scientist intern to develop Rider Agent simulations using state-of-the-art AI methods in a hands-on lab setting. You will build LLM-based agents that reflect diverse rider contexts and validate them against real data to accelerate product iterations.

This role blends LLM engineering, agent-based modeling, machine learning, and causal inference, with a focus on creating a working Rider Agent prototype and an evaluation framework for simulation fidelity.

Qualificações

• PhD candidate with expected graduation 2027–2028.
• Proficiency with Python in production environments.
• Hands-on experience with LLMs or agent-based systems.
• Strong ML knowledge and empirical model evaluation.
• Ability to independently develop prototypes.
• Strong written and verbal communication skills.
• Familiarity with A/B testing, causal inference, or experimental design.

Responsabilidades

• Develop LLM-based Rider Agents that represent heterogeneous rider contexts.
• Build agent-based simulation environments to test product experiences.
• Create evaluation pipelines for realism, robustness, and mechanism plausibility.
• Analyze emergent behaviors under different user segments and markets.
• Conduct experiments and ablation studies on agent behavior and simulation validity.
• Apply framework to real Rider problems for hypothesis generation.
• Communicate findings to science, engineering, and product teams.

Conhecimentos

Python

LLMs

Agent-based systems

Machine Learning

Formação

PhD in Computer Science / ML / AI / Data Science

Descrição da vaga

Lyft is seeking an Applied Scientist intern to develop Rider Agent simulations using state-of-the-art AI methods in a hands-on lab setting. You will build LLM-based agents that reflect diverse rider contexts and validate them against real data to accelerate product iterations.

This role blends LLM engineering, agent-based modeling, machine learning, and causal inference, with a focus on creating a working Rider Agent prototype and an evaluation framework for simulation fidelity.