Took ownership of enterprise-wide Azure AI Foundry platform operations for a regulated environment, covering full lifecycle management of LLM and embedding model deployments, workspace governance, and internal platform adoption across multiple product teams. Governance frameworks were aligned with EU AI Act, ISO 27001, and DORA requirements. Extended the platform with Semantic Kernel-based, vector database integration for RAG-based knowledge retrieval for internal process orchestration.

Responsibilities:

• Administered and maintained Azure AI Foundry workspaces across dev, test, and production environments, enforcing consistent workspace structure, compute configurations, and model catalog standards.
• Managed end-to-end LLM and embedding model deployment lifecycle within Azure AI Foundry, including version control, staged rollouts, deprecation workflows, and quota allocation across business units.
• Configured and operated Portkey as the AI gateway layer for model routing, request caching, semantic guardrails, and unified observability across AI endpoints.
• Integrated Azure AI Gateway for token-based rate limiting, quota enforcement, semantic caching, and load balancing across model deployments, with AI Content Safety integration for automated prompt moderation and compliance.
• Evaluated and integrated Semantic Kernel as an agent skill orchestration layer, enabling reusable, composable AI capabilities across internal product teams.
• Integrated vector databases for RAG-based knowledge retrieval, enabling semantic search across enterprise document repositories and internal knowledge bases.
• Implemented governance and compliance controls aligned with EU AI Act, ISO 27001, and DORA requirements, including audit logging, model lineage tracking, and explainability tooling for regulated AI deployments.
• Configured IAM-based access policies and enforced secure data flows and GDPR-compliant data handling across all Azure AI Foundry workspace components.
• Designed and implemented reusable AI agents and tool interfaces through MCPs, exposing internal services via APIs and MCP-compatible patterns.
• Built agent execution workflows, enabling LLM-driven systems to trigger and orchestrate real-world actions through structured service calls.
• Designed and implemented enterprise system integrations, connecting AI platforms with internal business services and external APIs to enable real-world data access and actions.

• Designed and implemented the core control-plane architecture, including an API Server with RBAC, Controller Manager, Scheduler-like reconciliation loops, Namespace isolation, and CRD-style resource definitions.
• Developed client SDKs and code-generation tools in Go and TypeScript to streamline custom controller development for internal engineering teams.
• Provisioned and managed AWS infrastructure using AWS CDK (infrastructure-as-code), deploying workloads across AWS Lambda (serverless event handlers), ECS (containerized services), and S3 (object storage backends).
• Implemented SQL/ORM-based resource state tracking and tenant isolation across the control plane data layer, supporting consistent reads and optimistic concurrency for thousands of concurrent tenants.
• Integrated WebSocket for real-time event streaming between control plane and worker nodes and utilized Redis for distributed caching and message brokering.
• Employed LocalStack for local AWS service emulation in CI/CD pipelines.
• Integrated Prometheus, Alertmanager, and Grafana for end-to-end metrics instrumentation, distributed tracing, and proactive anomaly detection.
• Implemented Kubernetes-native patterns: Deployments, StatefulSets, DaemonSets, custom operators.

• esigned and implemented a Cloud Foundry service broker (Node.js), enabling automated provisioning, binding, and lifecycle management of platform services.
• Architected and developed billing and metering microservices to capture, aggregate, and process resource usage data, forming the foundation for usage-based pricing and chargeback models.
• Designed event-driven and API-based communication between microservices, ensuring reliable data flow and consistency across billing pipelines.
• Built scalable data processing and storage layers (PostgreSQL, Redis) to handle high-volume usage metrics and support concurrent tenant workloads.
• Implemented end-to-end observability using Prometheus and Grafana, including metrics instrumentation, alerting, and dashboards to ensure system reliability and performance.
• Designed and exposed internal and external APIs to support seamless integration with platform services and customer-facing systems.
• Ensured horizontal scalability and resilience of services, supporting hundreds of concurrent tenants with consistent performance

• Architected and implemented infrastructure-as-code using Terraform and configuration management with Ansible to automate SAP HANA installation, upgrades, and system lifecycle operations across environments.
• Designed and developed an event-driven automation layer, including a lightweight agent integrated with Consul for service discovery and change propagation, triggering dynamic infrastructure workflows.
• Built and exposed APIs to handle customer HANA system provisioning and lifecycle operations, enabling self-service and programmatic access.
• Engineered secure, production-grade AWS network architecture (VPC, subnets, routing, IAM), ensuring isolation, compliance, and high availability across regions.
• Implemented automated backup and disaster recovery strategies leveraging S3 and Glacier, ensuring data durability and rapid restoration.
• Reduced end-to-end deployment time by 40% through automation and system optimization while maintaining strict enterprise security and compliance standards.

• Built cloud-native microservice following 12-factor app principles with stateless design, externalized configuration, and graceful shutdown
• Implemented multi-layer caching strategy using Redis (L1 in-memory cache with TTL/LRU eviction) and MongoDB (L2 persistent cache for complex query metadata)
• Designed cache key generation algorithm incorporating RBAC permissions, tenant isolation, and OLAP cube context (dimensions, measures, filters)
• Developed cache invalidation strategies with pub/sub patterns for real-time data updates
• Integrated Prometheus with custom metrics (cache hit/miss ratios, query latency percentiles, eviction rates)
• Built Grafana dashboards for real-time performance monitoring and capacity planning.