Senior MLOps Engineer

EPAM is a leading global provider of digital platform engineering and development services. We are committed to having a positive impact on our customers, our employees, and our communities. We embrace a dynamic and inclusive culture. Here you will collaborate with multi-national teams, contribute to a myriad of innovative projects that deliver the most creative and cutting-edge solutions, and have an opportunity to continuously learn and grow. No matter where you are located, you will join a dedicated, creative, and diverse community that will help you discover your fullest potential.
We are seeking a Senior MLOps Engineer to join an MVP engagement with a major AAA game publisher, building a test intelligence platform for two game franchises in parallel. A core design principle is full re-derivability and model lineage from day one — every run must be replayable from its stored configuration version and feed read positions. The signal catalog feeds a scoring strategy engine with versioned configurations, and calibration sweeps over historical data produce suggested weight updates surfaced directly in the Settings View.
This role ensures the ML and signal components are production-ready, reproducible, and improvable over time, forming the foundation of the system's long-term value as franchise history accumulates and models are refined.
Responsibilities
Own Signal Catalogue operations: signal refresh orchestration triggered by feed read-position advances, grain translation between per-test, per-area, and per-run signal families, and provenance capture across all 8 signals
Operate the Semantic Vector Index versioning: coordinate with the Senior AI Developer on model and dimension stamp conventions; design and execute the controlled reindex path when the enterprise AI gateway model changes
Design, implement, and own the Back-test & Calibration Harness: as-of temporal filtering across all record families, replay runner, look-ahead spot audit, and configuration sweep runner
Enforce holdout patch-set discipline, configuration sweep over route limits, thresholds, weights, and Composition setting; produce catch-rate vs. scope tables per candidate configuration and publish winning configurations as suggested-weight proposals into the Settings View
Lead Model Generation & Experimentation: systematic experimentation framework over scoring strategy configurations, tracking which signal weights and route combinations yield the best catch-rate vs. scope trade-off
Maintain model lineage across configuration versions for both franchises
Implement the MLOps Monitor and Data-Health Monitor: catch-rate floor monitoring, run-behaviour drift counters (per-run candidate volumes per route, score distribution vs. usual range), data-health telemetry across all ingestion channels
Manage exploration cadence support: unbiased random-sample injection with provenance ensuring exploration entries are never counted as model recommendations
Contribute to operator runbook sections covering signal refresh, calibration campaigns, model generation runs, and embedding reindex procedures
Requirements
3+ years of experience in MLOps or ML platform engineering
Expertise in ML model lifecycle management, including versioning, configuration management, and rollback
Background in signal computation pipeline design, covering scheduled refresh, provenance capture, and grain translation
Proficiency in calibration methodology: holdout discipline, configuration sweep design, and catch-rate vs. scope measurement
Knowledge of as-of temporal data systems or back-test harness design and operation
Skills in Python and SQL for ML pipeline automation
Competency in model monitoring, including drift detection, catch-rate floor monitoring, and run-behavior drift counters
Capability to collaborate with data engineers and AI developers on feature alignment
Qualifications in documenting calibration procedures, signal definitions, and operational runbooks
Familiarity with Spec Driven Development
English proficiency at an Upper-Intermediate level (B2) or higher
Nice to have
Understanding of MLflow, Kubeflow, or equivalent experiment tracking platforms
Familiarity with Snowflake ML or Snowpark
Showcase of embedding model versioning and controlled reindex orchestration
Skills in pgvector or vector store operational management
Background in gaming domain or QA tooling
We offer
International projects with top brands
Work with global teams of highly skilled, diverse peers
Healthcare benefits
Employee financial programs
Paid time off and sick leave
Upskilling, reskilling and certification courses
Unlimited access to the LinkedIn Learning library and 22,000+ courses
Global career opportunities
Volunteer and community involvement opportunities
EPAM Employee Groups
Award-winning culture recognized by Glassdoor, Newsweek and LinkedIn
EPAM is an Equal Opportunity Employer. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, national origin, age, sexual orientation, gender identity or expression, disability, protected veteran status, or any other characteristic protected by applicable law.

Work arrangement
Yes