MLOps Engineer IT & Data Development

Responsibilities

• Design, build, and maintain an ML platform that automates the entire model lifecycle, from training and validation to deployment, monitoring, and retraining
• Develop AutoML infrastructure to automate experiments, hyperparameter tuning, model evaluation, and promotion from staging to production
• Implement and maintain MLflow or similar tools for experiment tracking, model versioning, and reproducibility
• Design and maintain a Feature Store with online and offline storage, ensuring point-in-time correctness and feature consistency between training and inference
• Automate training, retraining, and scoring pipelines using Airflow and other orchestration tools, supporting both batch scoring and real-time inference
• Build CI/CD pipelines for ML, including automated data and model testing, quality checks, safe deployments, and rollback mechanisms
• Implement production monitoring for ML systems, covering availability, latency, throughput, resource utilization, data drift, concept drift, and model performance
• Containerize, deploy, and scale ML services using Docker and Kubernetes, ensuring reliability and fault tolerance
• Develop Infrastructure as Code (IaC) practices and manage environments, configurations, secrets, and access using Terraform, Ansible, Vault, and similar tools
• Work closely with Data Engineering to integrate the ML platform with ClickHouse, Kafka, PostgreSQL, Redis, and other data infrastructure components
• Establish MLOps standards, reusable pipeline templates, and tooling that streamline model development, testing, and operations for Data Science and ML Engineering teams

LLMOps & LLM Infrastructure

• Build and maintain infrastructure for deploying and operating LLM services, including environment setup, configuration management, and model versioning
• Configure and optimize LLM inference using vLLM, Ollama, and similar tools, focusing on GPU utilization, batching, latency, throughput, and memory efficiency
• Provide unified API access to multiple LLMs through LiteLLM or equivalent solutions, implementing request routing, fallback mechanisms, and usage controls
• Integrate LLM services with internal applications and platforms, including Onyx
• Automate LLM service deployments and updates, ensuring reproducible configurations, model versioning, and controlled releases
• Implement LLM infrastructure monitoring and optimize inference performance, operating costs, scalability, and reliability
• Integrate LLM services with existing ML pipelines and platform tooling when needed

Requirements

• Hands-on experience building or operating MLOps infrastructure or ML platforms
• Strong Python and Linux skills, including infrastructure automation and development of supporting services
• Experience with Docker and a solid understanding of Kubernetes and container orchestration
• Experience building CI/CD pipelines and managing infrastructure using Terraform, Ansible, or similar tools
• Solid understanding of the full ML model lifecycle, including training, validation, registration, deployment, monitoring, and retraining
• Experience with MLflow, Airflow, Feature Stores, or comparable technologies
• Experience implementing monitoring, logging, alerting, and troubleshooting for production systems
• Understanding of batch scoring and real-time inference architectures
• Experience integrating services with message brokers and analytical data stores
• Experience building AutoML platforms and automating experiments using Optuna or Ray Tune
• Hands-on experience deploying and operating LLMs or building LLMOps infrastructure
• Experience with vLLM, Ollama, LiteLLM, Ray, or similar tools
• Experience managing GPU infrastructure and optimizing LLM inference performance
• Experience with Feast or other Feature Store solutions
• Experience monitoring data drift, concept drift, and ML model performance in production
• The opportunity to work on large-scale products and contribute to their development
• A flexible work environment built on autonomy, ownership, and trust, without micromanagement
• The option to work fully remotely or from a modern office
• An annual $2,000 benefits allowance after the probation period, which can be used for health insurance upgrades, language learning, and sports activities
• Partial reimbursement for professional training, courses, and participation in industry conferences
• Access to a corporate digital library
• Company-funded weekly online sessions with a psychologist
• Relocation opportunities to countries where the company has offices, with relocation assistance available after the probation period, subject to approval