Want to know if this job is worth applying to?
United States
Remote only
OpenRouter is the leading AI routing and infrastructure layer that developers and enterprises use to access, manage, and optimize the best large language models across providers without lock-in, capacity constraints, or unnecessary cost. We power the most advanced AI teams in the world by giving them the flexibility to move fast, scale confidently, and stay future-proof as models evolve.
As enterprise adoption of AI accelerates, OpenRouter sits at the center of how organizations operationalize LLMs across research, product, and production workloads.
As a Data Scientist, you will build the intelligence layer that powers how millions of developers and AI agents interact with the world's largest model marketplace. OpenRouter processes billions of LLM generations spanning every major model, provider, and use case. Your job is to turn that data into a better product and a better experience for our customers.
You will work directly with our customer-facing and product teams to understand how requests are routed, where customers hit friction, which behaviors drive growth or churn, and what insights we should surface back to users so they can understand and optimize their own AI usage.
This is not a reporting role. You will own questions end to end: framing the problem, digging into the data, and shipping the answer. That might be a change to our routing heuristics, a new analytics product for customers, or a recommendation that shapes the roadmap. Whatever form it takes, the insights you generate feed directly back into the product, and you will see your work in the experience of every customer we serve.
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.
Experience & Technical 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.
Mindset & Approach
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.
Bonus Points
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.
