MLOps Engineer (Remote)

Project Details: Own the end-to-end lifecycle of production ML: training, packaging, deployment, monitoring, and governance. Build reusable pipelines and tooling so data scientists and contractors can ship reliable model quickly - batch and real-time - on Google cloud. 
 
Must Have Skills:

• 4+ years of MLOps/ML platform or DevOps for data/ML systems
• Hands on GCP experience: BigQuery, Cloud Run, Cloud Storage, Pub/Sub, Cloud Build (Vertex AI a plus)
• Proficiency with Python, packaging (Docker), and CI/CD 
• Solid SQL skills and understanding of data modeling for ML features/labels 
• Experience operating production models with monitoring, alerting, and incident response 

 
Soft Skills:
Nice to have Skills:

• Model registry & experiment tracking (ML Flow, W&B, or Vertex AI)
• Data validation & monitoring (Great Expectations, TensorFlow Data Validation, WhyLabs, Arize) 
• Feature store concepts (BQ-based or managed) 
• Canary/shadow deployments, autoscaling, and performance tuning 
• IaC (Terraform), testing frameworks (unit/integration/lead), and observability (Open Telemetry, Cloud Monitoring) 

Education/certification requirements:

• N/A

Day to Day responsibilities:

• Pipelines & orchestration: Design CI/CD and scheduled pipelines for training and inference (Cloud Build, Workflows/Scheduler, Pub/Sub, Cloud Run; Vertex Pipelines if used).
• Packaging & deployment: Standardize model packaging (Docker), artifact/versioning, and rollout strategies (A/B, canary, shadow) with automated rollbacks.
• Data/feature flows: Define contracts for features/labels in BigQuery and manage backfills; support batch and (where applicable) streaming features.
• Registry & experimentation: Stand up a model registry and experiment tracking (MLflow/Weights & Biases/Vertex) with approvals and audit trails.
• Monitoring & quality: Implement data/feature validation, drift/decay monitoring, performance/latency SLOs, and alerting; build dashboards and playbooks.
• Security & compliance: Enforce IAM least privilege, service accounts, Secrets Manager, provenance/lineage, and change management.
• Cost & performance: Track training/inference cost and latency; optimize hardware/ autoscaling and query patterns.
• Enablement: Create templates, docs, and tooling so DS/contractors can add models with minimal friction.

Tech stack you’ll use

• Compute/Orchestration: Cloud Run, Workflows/Scheduler, Pub/Sub, Vertex Pipelines (optional)
• Data/Storage: BigQuery, Cloud Storage (artifacts, datasets)
• CI/CD & IaC: Cloud Build or GitHub Actions, Terraform
• ML Tooling: MLflow/W&B/Vertex, Docker, PyTorch/TF/XGBoost (as provided by DS)
• Monitoring: Cloud Logging/Monitoring, Evidently/WhyLabs/Arize, custom run IDs & metrics

How we work

• Small, versioned releases; test-first pipelines; documented runbooks.
• Clear SLOs and blameless incident reviews.