Note: The job is a remote job and is open to candidates in USA. OpenRouter is an AI routing and infrastructure layer that helps developers and enterprises access, manage, and optimize large language models across providers. The Data Scientist will build routing models, analytics products, and user-facing data features while analyzing large-scale LLM usage to improve the product and inform strategy. The role also involves defining product metrics and contributing to the scalable data warehouse.
Responsibilities
- Build and improve the rules, heuristics, and models that drive our core model routing and provider routing systems
- Build analytics products and conduct deep-dive analyses that empower users to understand usage, diagnose issues, and optimize workflows while informing our product strategy
- Design and ship user-facing data features, from task-type classification taxonomies to public rankings pages that break down model performance by use case
- Define and track product metrics such as feature adoption, user engagement, and platform health, directly tying these insights to roadmap and architectural decisions
- Contribute meaningfully to the data warehouse (dbt, ClickHouse) to ensure the analytical foundation scales with the business
Skills
- • Experience building and shipping production-level data products such as models, algorithms, classifiers, scoring systems, or automated processes
- • Proven ability to partner with engineering and product teams to convert insights and analytics into customer facing features
- • Strong SQL skills. You can write complex, performant queries against large-scale analytical databases (ClickHouse, BigQuery, or similar)
- • Deep fluency with AI and agents. You use LLMs and AI tools in your daily work, understand how they behave, and have informed opinions on model evaluation, prompt engineering, and the agentic ecosystem
- • Excellent statistical and mathematical foundations. You think in distributions, causal inference, and experimental design, not just averages and dashboards
- • Proficiency in Python (or similar scripting language). You can prototype models, run simulations, and build data pipelines
- • 4+ years of relevant experience in a data science, applied ML, or quantitative product role at a high-growth, high-velocity company
- • Product-oriented. You build for the developer, not just for intellectual interest. You have strong product sense and can translate ambiguous, complex systems into clear, shippable solutions
- • Highly motivated and independent. You thrive in fast-moving, low-structure environments. You don't wait for tickets; you identify the highest-leverage problem and go solve it
- • AI-first in your own workflow. You use coding agents, MCP servers, and modern AI tools heavily and have strong opinions about what works
- • Incredible problem solver. You can navigate ambiguity, break down complex systems, and find creative approaches where others see dead ends
- • Strong written communication. You can distill complex analysis into clear recommendations for technical and non-technical audiences alike
- • Experience with LLM evaluation and benchmarking such as online/offline evals, human-in-the-loop assessment, safety/quality guardrails
- • Experience with developer platforms, APIs/SDKs, or usage-based products
- • NLP or machine learning engineering experience building classifiers, embeddings, or ranking models
- • TypeScript experience. Parts of our stack (routing engine, API layer, front-end) are TypeScript. Being able to contribute directly to the product codebase is a meaningful plus
Company Overview