Deekshith Chandupatla
Professional Summary
Software Engineer with 2+ years of experience building backend systems, high-throughput event pipelines, and experimentation platforms. Strong hands-on experience with Java and Spring Boot for microservices, Apache Kafka and Spark for streaming, and AWS-hosted PostgreSQL for high-throughput data stores. Experienced containerizing and running services with Docker and Kubernetes, and automating delivery with GitHub Actions and Jenkins. Implemented caching with Redis, tracing with OpenTelemetry, and metrics collection with Prometheus to improve operability of low-latency services. Integrated ML-driven RAG and embedding-based retrieval using OpenAI APIs and pgvector, and built Python/FastAPI prototypes for production retrieval. Seeking backend/platform roles focused on distributed systems, experimentation, and ML-enabled pipelines.
Technical Skills
Work Experience
- Architected and implemented Spring Boot microservices that expose REST and gRPC APIs to serve experimentation workflows and feature evaluation.
- Optimized Redis caching policies and request handling to reduce tail latency, enabling the experimentation platform to sustain 3K+ TPS.
- Tuned AWS Aurora PostgreSQL schemas, queries and partitioning and deployed PgBouncer and read replicas to improve query performance by 40%.
- Engineered Apache Kafka event streams and developed Apache Spark processing jobs to decouple producers and consumers for experimentation events.
- Containerized services with Docker and deployed them to Kubernetes to standardize rollouts and simplify scaling.
- Built CI/CD pipelines using GitHub Actions and Jenkins to automate builds and shorten release cycles.
- Integrated AWS Lambda functions to handle asynchronous experimentation tasks and offload heavy processing.
- Implemented OpenTelemetry tracing and Prometheus metrics collection to reduce incident detection time to under 5 minutes for high-throughput services.
Projects
- Built a production-grade RAG agent using Python and FastAPI that exposes REST endpoints and uses PostgreSQL with pgvector for semantic retrieval, achieving ~150ms average retrieval latency across 50K+ embedded records.
- Implemented token accounting and LLM cost-tracking in PostgreSQL to create auditable per-request usage logs and reduced LLM cost overruns by 30% in multi-user workloads.
Education
Powered by Drivetube · Create your own profile at drivetube.ai