I spent three years at Microsoft keeping Azure backup and restore dependable for enterprise workloads, and now build retrieval-augmented AI systems at NYU while finishing my Masters. I like infrastructure that is fast, fun to operate, and occasionally shaped like a Game Boy game.
Deployed retrieval-augmented generation (RAG) systems for the Pilot GenAI initiative using LangChain and pgvector, lifting retrieval precision by 12% across enterprise knowledge bases.
Integrated fine-tuned LLMs (GPT-4o, Claude 3.5, Llama 3) into core SaaS applications through FastAPI and Node.js services, sustaining sub-300ms initial response latency for teams across multiple university departments.
Implemented semantic caching in Redis for recurrent prompt queries, trimming API token spend and improving response latency by 60%.
Software Engineer
Jul 2022 - Jul 2025
Microsoft
Hyderabad, India
Shipped backup and restore orchestration features for Azure BCDR services, protecting SQL and SAP workloads on Azure VMs across large-scale production environments.
Devised a secondary job-validation pipeline that eliminated false restore failures, raising restore SLA compliance to 99.8% and preventing inconsistent state propagation across distributed backup services.
Engineered an automated machine learning remediation agent that resolved 800+ security vulnerabilities across 120+ repositories, reducing manual triage effort by 95% and strengthening compliance readiness.
Led high-availability backup support for SAP ASE, giving enterprise customers 24/7 resilient recovery, and migrated 80+ deprecated BCDR pipelines onto modernized infrastructure aligned with secure SDLC standards.
Directed recovery during a ransomware incident affecting 80+ customer VMs, designing retention-bypass workflows and sequencing restores to achieve full workload recovery within strict operational timelines.
Software Engineer Intern
Jun 2021 - Sep 2021, Jan 2022 - Jun 2022
Microsoft
Hyderabad, India
Delivered Partial Restore for Azure Backup, shortening incremental restore duration by 40%, and parallelized Full Sync execution to lower microservice load and operational cost by 25%.
Contextly
Apr 2026
Local RAG Application
A document-grounded question answering application that parses PDF and text files and returns answers with citations pointing to exact source passages, with the full RAG pipeline running locally.
Application is backed by Redis and MinIO S3-compatible object storage, with nomic-embed-text embeddings and an ms-marco-MiniLM-L-6-v2 cross-encoder reranker.
Next.js
TypeScript
FastAPI
PostgreSQL
pgvector
Celery
Erica AI
Sep 2025
GraphRAG Course Tutor
A Graph-based RAG course tutor with a multi-stage pipeline for ingestion, knowledge graph construction, and scaffolded answer generation spanning 1300+ RAG concepts.
Runs with a hybrid storage layer combining MongoDB for multi-source ingestion with Neo4j knowledge graphs modeling prerequisite relationships, with local GPU inference for privacy-preserving answers with structured citations.
Python
Neo4j
MongoDB
Ollama
GraphRAG
LinkedIn Post Generator
Dec 2025
AI writing assistant
An auto-post generator that drafts posts with Gemini or OpenAI, suggests source articles for inspiration, and tracks how each published post performed.
Allows user management, and post scheduling, with a React frontend and an Express backend using Drizzle ORM for PostgreSQL.
React
Express
Drizzle
PostgreSQL
Gemini
Education
New York University
Master of Science in Computer Science · Sep 2025 - May 2027 · New York, NY
BITS Pilani, K. K. Birla Goa Campus
Bachelor of Engineering in Computer Science · Aug 2018 - Jun 2022 · Goa, India
