Lead MLOps Specialist in ATB-market

We invite you to join our team Lead MLOps

Responsibilities:

Turn DS development into reliable services with predictable performance, transparent monitoring and secure releases (central and edge-levels).

Areas of responsibility: 

• ML-platform: registry of artifacts/models, version/access policies, service templates; 
• CI/CD for models: data/model tests, canary/blue-green/shadow, rollback, feature-flags; 
• Observability: quality/drift/stability, p95 latency/resources, incidents, feedback loops in training; 
• Security: secrets/IAM/RBAC, inference audit, config management, network policies/ingress; 
• Edge-scenarios: synchronization of models/caches, resistance to failures/outages, telemetry; 
• CPU/GPU capacity planning, error budgets for peak windows (evenings/Fridays); 
• Automated replay of inference logs for audit/retraining; zero-downtime updates; 

OKR examples:

1. Uptime of ML-services 99.5%; p95 latency <150 ms on critical paths; 
2. TtM model from approve to production <30 min; 95% of releases - without downtime; 
3. Automatic detection of data/model drift with a frequency of 24 hours; 

Requirements (must-have): 

• 5+ years in MLOps/SRE/DevOps; industrial exploitation of ML-services on-prem;
• Deep understanding of life cycle of models, risks and observability; 
• Confident knowledge of Kubernetes/OpenShift, Helm, Argo CD/Workflows, Terraform/Ansible, GitLab CI; 
• Production experience with MLflow Registry/Serving, NVIDIA Triton, ONNX Runtime, FastAPI/gRPC, KServe or Seldon Core; 
• Monitoring/logging: Prometheus/Grafana/Loki, Alertmanager, Evidently/whylogs, OpenTelemetry; 
• Security/configs: Vault/Sealed Secrets, Keycloak (IAM), CNI policies, ingress (Traefik/Kong/Nginx);
• Automation of data/model tests, incident management, runbooks.

Will be a plus:

• Edge inference in retail (POS/SCO/video/planograms); GPU-profiling, TensorRT/quantization/batch-policy;
• Multiversion models with fast roll-forward/back; cost-/energy-aware planning;
• Practice of cost-aware planning of resources and energy efficiency; 

The company offers:

• remote or hybrid formt of work;
• employment on the terms of a gig contract or in the state (reservation is possible);
• paid annual leave of 24 calendar days, paid sick leave;
• regular payment of wages without delays and in volumes, regular salary review;
• opportunity for professional and career growth;
• training courses.

Contact person: Kateryna, tel. style="font-weight: 400">0984567857 (t.me/KaterynaB_HR)