Lead the development and optimization of an AI Auto Router to predict model quality, cost, and latency. Own the end-to-end ML lifecycle including training data, objective functions, and production-parity evaluations. Requires 5+ years of experience in applied ML with a strong track record of shipping real-time decision models.
Key Highlights
Own the routing model end-to-end, treating it as a product rather than a research artifact
Develop rigorous evaluation frameworks to gate production changes using error bars and holdouts
Optimize for a complex objective involving task success, token cost, and wall-clock speed
Key Responsibilities
Own the routing model end to end, including features, objective, training, calibration, and release cadence.
Build production-parity benchmarks and router-vs-router comparisons to gate every change.
Convert fleet data, benchmark rollouts, and override feedback into training signals.
Model the user-experienced objective considering task success, token cost, and wall-clock speed.
Manage the online loop including shadow deployments, A/B testing, full rollout, and drift monitoring.
Experiment with calibrated decisions and abstention under uncertainty to push the routing frontier.
Partner with platform engineers on instrumentation for request-level features and labels.
Interested in remote work opportunities in Development & Programming? Discover Development & Programming Remote Jobs featuring exclusive positions from top companies that offer flexible work arrangements.
Browse our curated collection of remote jobs across all categories and industries, featuring positions from top companies worldwide.
Nice to Have
Experience with LLM routing specifically, including RouterBench-class evals and Pareto fronts
Experience with calibration, abstention, or conformal methods in production decision systems
Experience running online A/B loops with drift monitoring on models involving financial transactions
