Senior Cloud Architect & Generative AI Platform Engineer with over 15+ years of experience architecting large-scale data and GenAI infrastructure for enterprise production systems. I specialize in Generative AI (GenAI), LLMOps, MLOps, multi-agent frameworks, and building resilient, cost-effective cloud-native AI platforms.
- 🏢 Most Recent Role: Senior Cloud Architect & AI/ML Platform Lead at 7-Eleven, Inc.
- 🤖 Core Focus: Scaling Generative AI infrastructure and production-grade LLM applications across multi-cloud environments (Azure, AWS).
- 🚀 GenAI & Agentic Niche: Advanced RAG architectures, autonomous AI agents, enterprise LLM gateways, and multi-tenant GPU orchestration.
- Core Systems: OpenAI
GPT-4| AnthropicClaude| Hugging FaceTransformers - Frameworks: LangChain | LangGraph | LlamaIndex | Multi-Agent Frameworks (CrewAI, AutoGen)
- LLMOps Infrastructure: Databricks (Unity Catalog) | MLflow | vLLM Inference Optimization | Model Registries
- Containerization: Kubernetes (
AKS/EKSMulti-Tenant GPU Scheduling) | Docker | Helm - Pipelines & IaC: Apache Airflow (Astronomer) | Terraform | Ansible | GitLab CI/CD
- Cloud Architecture: Amazon Web Services (
AWS) | MicrosoftAzure| Google Cloud (GCP)
- Vector Search: Pinecone | Weaviate | Milvus |
pgvector(Hybrid Search / Semantic Caching) - Data Scale: Apache Spark | Databricks Delta Lake | Snowflake | Apache Kafka
- Data Reliability: Monte Carlo (Data-Quality Circuit Breakers)
- Infrastructure APM: New Relic | Datadog | Splunk | OpenTelemetry | Prometheus | Grafana
An autonomous GenAI troubleshooting agent built to diagnose live Kubernetes cluster issues. It leverages an advanced ReAct loop, Python-driven evaluation engines, and secure, read-only kubectl tool integrations to trace system faults dynamically via LLM reasoning.
A highly scalable Retrieval-Augmented Generation (RAG) assistant trained across the entirety of Kubernetes documentation. Fully containerized and deployed on K8s via Helm, utilizing pgvector and Ollama alongside automated performance, grounding, and accuracy evaluation cycles.
An enterprise-optimized layout for hosting large open-weights foundation models using vLLM on Kubernetes. Features advanced GPU multi-tenant scheduling setups, custom Helm charts, and metrics engines designed to optimize throughput and cost-per-token GenAI serving benchmarks.
- 7-Eleven, Inc. | Senior Cloud Architect & AI/ML Platform Lead
- Enterprise GenAI Leadership: Defined and drove the enterprise data and GenAI infrastructure roadmap powering retail intelligence across 13,000+ store locations.
- Production LLMOps & Architecture: Designed enterprise-grade RAG patterns using LangChain, LangGraph, and Azure OpenAI, including a secure corporate LLM API gateway equipped with prompt management, rate-limiting, and cost controls.
- GenAI Safety & Governance: Implemented multi-cloud foundation model access (Bedrock/Azure OpenAI) with safety guardrails to prevent prompt injections and secure customer PII; integrated Model Context Protocol (MCP) connectors linking AI coding environments directly to data catalogs for AI-assisted governance.
- GPU Infrastructure Management: Managed multi-tenant GPU/Kubernetes infrastructure (AKS/EKS, Azure NC/ND, AWS P3/G5) with auto-scaling, reducing compute overhead via quantization and Triton/vLLM optimizations.
- Apple | Senior DevOps Engineer - Infrastructure & Platform Engineering
- Containerized core internal ecosystems using Docker & Kubernetes for zero-drift deployments.
- Engineered automated validation and rollback CI/CD gates using Jenkins and Git across large-scale AWS data lakes supporting model artifact storage.
- CKA: Certified Kubernetes Administrator (The Linux Foundation)
- AWS: Certified SysOps Administrator – Associate & Certified DevOps Engineer – Professional
- HashiCorp: Terraform Associate
- Education: B.Tech. in Computer Science and Engineering — Pondicherry Engineering College, India
- LinkedIn: linkedin.com/in/sundar-djeabalane
- GitHub: github.com/arthuersundar
- Email: arthuer.sundar@gmail.com