Azure ML / AI Architect (Solution Engineering)

Azure ML / AI Architect
(Solution Engineering)

We’re seeking an Azure ML/AI
Architect who can design, build, and ship production AI solutions on Azure—then
partner with customers and field teams to land them. You’ll lead end -to -end
model lifecycle on Azure Machine Learning and Azure AI Foundry (Azure AI
Studio), orchestrate robust MLOps pipelines, and translate business goals into
scalable architectures. You’re equally comfortable whiteboarding with
executives, pairing with engineers, and tuning latency/cost for real -world
workloads. Experience across other clouds (AWS, Google Cloud, Oracle) is a
plus—we meet customers where they are.

Responsibilities

• Architecture &
  Delivery
  
• Own reference
  architectures for classical ML and GenAI (RAG, fine -tuning, tool/use -case
  orchestration) on Azure ML + Azure AI Foundry.
  
• Design secure, scalable
  MLOps with AML v2 (pipelines, components), GitHub Actions/Azure DevOps,
  model/feature registries, online/batch endpoints, and CI/CD.
  
• Build data/feature
  pipelines using Fabric/Synapse/Databricks, Delta/Parquet, and govern with
  Purview; integrate Key Vault, Private Link, VNets, Managed Identity.
  
• Productionize inference
  on Managed Online/Batch Endpoints or AKS; implement monitoring (drift,
  data quality, performance, cost) and A/B/Canary rollouts.
  
• GenAI & Apps
  
• Implement Azure
  OpenAI / Azure AI model catalog patterns (Prompt Flow, safety
  filters, content moderation, grounding with vector search).
  
• Deliver RAG systems
  (Azure Cognitive Search or vector DBs), retrieval evaluators,
  prompt/version management, and cost/latency optimization.
  
• Solution Engineering
  
• Lead discovery, write
  Solution/Architecture Design Docs, demo/reference apps, and run customer
  workshops/POVs.
  
• Partner with
  Sales/Customer Success; create estimates, landing zones, and handoffs to
  customer/managed services teams.
  
• Standards &
  Governance
  
• Embed Responsible AI
  practices (privacy, safety, fairness, transparency), threat modeling, and
  compliance -by -design.
  
• Establish coding
  standards, repo strategy, IaC (Bicep/Terraform), observability (App
  Insights/Log Analytics), and SRE runbooks.
  

Requirements

Required Qualifications

• 7–10+ years in data/ML/AI engineering with 3+ years
  building production solutions on Azure ML and Azure
  AI Foundry/Studio.
  
• Proven delivery of ML/GenAI projects end -to -end:
  problem framing, data/feature engineering, modeling, evaluation,
  deployment, and monitoring.
  
• Hands -on with: AML SDK v2 & pipelines, MLflow/Model
  Registry, Feature Store, Managed Endpoints/AKS, Prompt Flow, GitHub
  Actions/Azure DevOps.
  
• Strong Python engineering (PyTorch/Transformers or
  scikit -learn/lightGBM), containerization (Docker), and API design
  (FastAPI).
  
• Data platforms: Fabric/Synapse/Databricks; storage
  (ADLS, Delta); messaging/streaming (Event Hub/Kafka) fundamentals.
  
• Security & networking on Azure: Key Vault, Private
  Link, VNet, Managed Identity, RBAC.
  
• Executive -level communication; ability to lead
  architecture reviews and mentor engineers.
  

Preferred / Nice to Have

• Cross -cloud exposure: AWS SageMaker, Google
  Vertex AI, Oracle OCI Data Science / Generative AI; portability
  patterns across providers.
  
• Vector databases (Azure AI Search vector, Pinecone,
  Redis, pgvector), LlamaIndex/LangChain, evaluation frameworks (Ragas,
  Promptflow eval).
  
• Databricks (Unity Catalog, Feature Store), Power
  BI/Fabric Real -Time Intelligence, or Snowflake/Mosaic AI familiarity.
  
• IaC (Terraform/Bicep), Kubernetes (AKS), GPU workload
  tuning, Triton/ONNX, quantization/LoRA/SFT pipelines.
  
• Certifications: Azure AI Engineer/Architect;
  AWS/GCP/Oracle equivalents.
  

How You’ll Measure Success

• Production launches with measurable business impact (quality,
  latency, reliability, cost).
  
• Reusable assets: reference architectures, accelerators,
  and well -documented repos customers adopt.
  
• Clear governance & Responsible AI controls; zero
  critical security findings in reviews.
  
• Field enablement: workshops/POVs that convert to
  deployments.