ML DevOps

Prohibited locations: RF, Ukraine, RB
up to 5500 usd
English: B2+
Years: ML — 6+ months, DevOps -3+ years commercial experience

Existing team
·  FE, BE, BA, PM, UX/UI

Key Responsibilities
● Development and support of end-to-end ML pipelines (training, validation, deployment, monitoring, retraining)
● Construction and operation of CI/CD for models (test automation, packaging, and deployment)
● Design of LLM/RAG pipelines, context management,
embedding dashboards (embedding quality/dynamics dashboards), index regeneration, prompt and fact-check testing (Grounding/citation)
● MLOps platform setup: experiment tracking, model registry, feature store, monitoring
● Management of ML infrastructure and environments (GPU/CPU pools, Kubernetes/EKS, Docker)
● Implementation of deployment strategies: canary, shadow, A/B testing
● Ensuring model quality monitoring (accuracy drift, data drift, PSI, SLO/SLA)
● Artifact management (data, models, metadata, versions)
● Security compliance (encryption, access control, auditing, operation in private VPCs)
● Integrating ML models into backend services (API, gRPC, REST)
● Collaborating with Data Engineering and Data Science teams
● Documenting processes and best practices for ML infrastructure
● Managing the cost and scaling of ML infrastructure in AWS
● Data governance: storage policies (S3 lifecycle), dataset versioning (DVC/LakeFS), data lineage (OpenLineage), quality gates in CI/CD

Requirements
-ML Ops Tools
● MLflow or Kubeflow (experiments, registry)
● Feature Store (Feast, Tecton, or custom)
● Airflow, Prefect, or Kubeflow Pipelines (ML workflow orchestration)

-Infrastructure and Containerization
● Docker, Kubernetes/EKS
● AWS S3, ECR, EKS, IAM, KMS, VPC
● Terraform or Pulumi (IaC)
● GitHub Actions, GitLab CI, or Jenkins (CI/CD)
● Autoscaling, AWS Batch/Step Functions for offline processing and retrieval

-Monitoring and Observability
● Prometheus, Grafana, CloudWatch, CloudTrail
● Model Quality Metrics (AUC, F1, Brier, logloss)
● Stability metrics (drift detection, PSI)
● LLM-specific metrics: tokens/sec, context length, prompt/response size,
grounding rate, citation coverage, hallucination rate.

Key Competencies
● Building a stable and secure ML infrastructure
● Automation Full-cycle ML: from data to inference services
● Quality control and stability of models in production
● Effective collaboration with data science and data engineering teams

Joining Valletta Software Development means:

🌍 A Global, Thriving Team
Join 100+ specialists from 20+ countries, united by a passion for outstanding
IT solutions.
🚀Diverse projects: Fintech, MedTech, AI/ML, e-commerce, and more. Switch
teams or industries to broaden your skills.
💡 Support at Every Step Client interview prep: We train you to succeed + give actionable feedback.
✔️ Strategic stability: Well-structured processes, strong management, and long- term vision.
✔️ Core values: Honesty, flexibility, innovation, and a people-first approach.
💸 Regular salary review based on your personal results
✨ Paid rest days and sick leaves;

Фильтры: Middle, Senior, Удаленная работа