Hemanth Janapala

Reference-data pipeline migration

I led the end-to-end migration of a core promotion-targeting pipeline from legacy Table Storage and Databricks to an event-driven SQL architecture that processes millions of records daily.

• Built a Spring Boot batching engine with idempotency, retries and fault-tolerant ingestion.
• Tuned indexes, queries and JDBC batching to hold throughput across millions of records per run.
• Reached 99.9%+ delivery to downstream systems, reprocessing unacknowledged events within a 5-minute SLA.

• Spring Boot
• Azure Data Factory
• Event Hub
• SQL Server
• JDBC

Promotion audit service

I led the design and delivery of an internal audit service with a dual-store persistence layer: Cosmos DB for structured audit records and OpenSearch for fast label retrieval.

• Implemented idempotent event handling with dead-letter queues so no audit record is ever duplicated.
• Designed OpenSearch index mappings and query optimizations for high-throughput lookups.
• Set up Azure Monitor and Application Insights alerts on consumption rate, latency and write throughput.

• Cosmos DB
• OpenSearch
• DLQ
• Application Insights

Platform security hardening

I replaced static, manually rotated credentials with Azure AD workload identity, and found and redesigned a vulnerable header-based service-to-service auth scheme using token-based verification.

• Azure AD Workload Identity
• OAuth2
• JWT

Conversational data agent

I contributed to an LLM-based agent built with LangGraph and Azure OpenAI that lets teams ask questions about promotion and audit data in plain language and validate the answers.

• LangGraph
• Azure OpenAI
• Python