Remote Senior Applied ML Engineer - Real-Time Model Router

Kimchi; Spain; Remote; EUR 90,000 - 140,000; Full time

Benefits offered by this job

Competitive salary

Remote-first global environment

Equity options

Learning budget

Annual hackathon

Job summary

Kimchi is building an Auto Router to decide which model serves each request, balancing quality and cost. This applied-ML role sits in a platform team that owns the proxy, streaming path, and fleet, while you own the model inside—training data, objectives, evaluation, and the loop that ships a better router.

You will work with Python, SQL at scale, and ML infra across a distributed team to define production benchmarks, calibrate decisions, and monitor drift, delivering real-world savings and

Qualifications

• 5+ years in applied ML or ML research with real-time decision model deployment.
• Strong Python and SQL at scale for big data analysis and production queries.
• Ability to distinguish between model improvements and evaluation ease.
• Form opinions on LLM benchmarks and agentic harnesses in production.
• Understand serving realities: TTFT, decode speed, batching, concurrency.
• Instrument before optimization, with error bars and tracked experiments.
• A wide mandate and curiosity about complex routing problems.

Responsibilities

• Own the routing model end to end: features, objective, training, calibration, and release cadence.
• Build production-parity benchmarks and router-vs-router comparisons.
• Turn fleet data into training signals with clear failure categorization.
• Model the user-experienced objective: task success × token cost × wall-clock speed.
• Run online loops: shadow, A/B, full rollout with drift monitoring and retraining.
• Push frontier with calibrated decisions and speed-aware routing.
• Collaborate with platform engineers on instrumentation and data-path support.

Skills

Applied ML

Real-time decision making

Python

SQL at scale

Experimental mindset

System design thinking

Tools

Python

SQL

RouterBench-class

GitLab CI

ClickHouse

Job description

Kimchi is building an Auto Router to decide which model serves each request, balancing quality and cost. This applied-ML role sits in a platform team that owns the proxy, streaming path, and fleet, while you own the model inside—training data, objectives, evaluation, and the loop that ships a better router.

You will work with Python, SQL at scale, and ML infra across a distributed team to define production benchmarks, calibrate decisions, and monitor drift, delivering real-world savings and