Senior Software Engineer with 10+ years building scalable data pipelines, cloud-native infrastructure, and enterprise backend systems. Currently working on indexing and searching petabytes of M365 data.

🌐 hiteshpattanayak.info · AWS Community Builder · CKAD Certified

Languages: Go · Python · TypeScript / Node.js · PySpark
Data Engineering: Databricks · Apache Spark · Delta Lake · Azure Event Hubs
Cloud & Infra: Kubernetes · Docker · Azure · AWS · Terraform · Pulumi
Databases: CosmosDB · PostgreSQL · Elasticsearch · TimescaleDB
AI / LLM: RAG pipelines · Azure OpenAI · Anthropic API · Vector Search
Protocols & APIs: gRPC · REST · GraphQL

• Ultimate CKAD Certification Guide — OrangeAva
• Modern API Design with gRPC — OrangeAva

• Flash talk — gRPC Load Balancing @ GopherCon 2023
• Virtual talk — Microservice Communication using gRPC @ AWS UG Bangalore
• Blog featured in kube-weekly
• Blog featured in LearnK8s LinkedIn pulse

• Semantic Search (RAG) — CosmosDB hybrid vector search + Azure OpenAI over petabytes of M365 backup data; natural language → metadata filters via few-shot Chat Completions
• Elastic Dashboard Changelog — Python + Anthropic API tool that diffs unreadable .ndjson Kibana files and generates human-readable changelogs
• Security Fix Automation — LLM-assisted local skill that ingests Cycode findings and applies targeted fixes with full code context
• Blog Generator — AI-powered workflow (Claude / OpenAI) to draft posts from structured idea files
• AI Chat Assistant — RAG conversational assistant on my blog site (TF-IDF + Netlify Functions + GPT-4o-mini)

My blog has a built-in AI chat assistant. Ask it about my posts, projects, or background — it retrieves relevant content and answers using GPT-4o-mini.

👉 Chat at hiteshpattanayak.info

I'm diving into the intricacies of enhancing retrieval-augmented generation (RAG) systems by optimizing vector databases using approximate nearest neighbor (ANN) algorithms. My focus is on improving the efficiency of chunking strategies to better manage and retrieve relevant information swiftly within dynamic content contexts, particularly for AI chat assistants. Additionally, I'm exploring how to seamlessly integrate per-page context to enhance user interactions without the heavy lift of maintaining a full vector database.

_{Powered by Claude via scheduled GitHub Actions · view workflow}