Senior Real-Time ML Platform Engineer

Embedded Shishya; San Francisco, New York, Portland (CA, NY, OR); Hybrid; USD 167,000 - 208,000; Full time

Job summary

Mercury in San Francisco, CA is seeking an experienced ML Engineer to build and operate the real-time inference service for our risk decisioning platform. You will own deployment infra, ensure low latency and high availability, and partner with Data Science to move models from development to production under MLP ownership.

The ideal candidate has 5+ years in ML engineering or MLOps, strong Python backend skills with FastAPI or Flask, and hands-on experience with model registries, CI/CD, and

Qualifications

• 5+ years in ML engineering, backend software, or MLOps.
• Production ML service experience in low-latency contexts.
• Strong Python backend skills; FastAPI or Flask.
• Experience deploying ML models: registries, CI/CD, versioning, staged rollouts.
• Observability and alerting for production services; drift signals.
• Familiar with SQL and low-latency stores; streaming pipelines (Kafka, Kinesis).

Responsibilities

• Build and operate real-time inference service with low latency and high availability.
• Own model deployment infra: registry, CI/CD, versioning, shadow/canary rollouts.
• Develop observability: latency, errors, drift detection.
• Collaborate with Data Science to bring models into production under MLP ownership.
• Enable experimentation like champion/challenger and canary routing.

Skills

Python

FastAPI

MLOps

CI/CD

Observability

Kafka

SQL

Tools

Redis

DynamoDB

Kinesis

Snowflake

Job description

Mercury in San Francisco, CA is seeking an experienced ML Engineer to build and operate the real-time inference service for our risk decisioning platform. You will own deployment infra, ensure low latency and high availability, and partner with Data Science to move models from development to production under MLP ownership.

The ideal candidate has 5+ years in ML engineering or MLOps, strong Python backend skills with FastAPI or Flask, and hands-on experience with model registries, CI/CD, and