Senior Applied ML Engineer | Kimchi (Model Router)

CAST AI is hiring a Senior Applied ML Engineer for Kimchi, its model router for LLM requests. The role focuses on predicting quality, cost, and decoding speed across candidate models, with the aim of preserving task success while reducing token costs. It owns the routing model’s training data, objective, evaluation, calibration, and release cycle.

The engineer will build production-parity evaluations, turn fleet and benchmark outcomes into training signals, and run shadow tests, A/B experiments, rollouts, and drift monitoring. The work also includes accounting for serving conditions such as concurrency and caching, and partnering with platform engineers on instrumentation. Candidates need at least five years in applied ML or research and experience shipping models for real-time decisions. The posting lists Italy among the eligible source locations and describes the environment as remote-first.

Highlights

• Own the router model end to end, from features and training to release.
• Build evaluations with task-family holdouts and error bars.
• Run online experiments and monitor production drift.
• Work with Python, PyTorch, SQL, and LLM APIs.