Durva Dev

MS at Penn State. Built RAG applications and maintained legacy banking software. Recently built agentic AI at the Amazon Nova Challenge.

Skills

Languages
C/C++ · Go · JavaScript · TypeScript · Python
Systems
Raft · gRPC · Kubernetes · Terraform · PostgreSQL · AWS · GCP
AI
RAG · Agents · MCP · Bedrock · LangChain · LangGraph

Projects

Experience

Agentic AI Developer

01/2026 – 05/2026

Amazon Nova AI Challenge @ PSU — State College, PA

  • Built a Python agentic orchestration system for code generation using Strands Agents, Amazon Bedrock, and Claude Agent SDK for repository analysis, SpecKit-style planning, and multi-turn coding-agent feedback.
  • Implemented LLM-as-a-Judge specification validation and execution-based code evaluation using GitHub, CodeBuild, Playwright, and JUnit to guide revisions and enforce test-based approval gates.
  • Engineered reliable AWS agent infrastructure with Lambda, DynamoDB/S3 state persistence, idempotent retries, deadline-aware execution, and protocol safeguards.

Systems Engineer

07/2022 – 07/2024

Tata Consultancy Services — Mumbai, India

  • Project MARG: Built and deployed a RAG application across 3K+ enterprise documents, enabling semantic retrieval of relevant business content and reducing sales-response turnaround from 3–4 weeks to 2–3 days.
  • Engineered the end-to-end knowledge pipeline for document ingestion, chunking, embeddings, metadata enrichment, vector indexing, and query-time retrieval, reducing search latency by 40%.
  • Project SNB: Developed regulatory-compliance and delinquent-loan reporting workflows using IBM MQ, while optimizing Oracle PL/SQL stored procedures across 11M+ records through execution-plan analysis to resolve production failures.

Education

Penn State University

08/2024 – 05/2026

M.S. Computer Science — University Park

University of Mumbai

08/2018 – 05/2022

B.E. Computer Engineering — India

Contact

Say hello. Both GitHub accounts are linked below.