ABOUT

I'm Arnav, a student at Boston University pursuing B.A's in both Computer Science and Economics. My main interest lies in distributed systems and my recent internship at Grepr sparking fascination with stateful, fault-tolerant stream processing engines. I also like thinking about scaling AI/ML systems in a world where compute continues to become increasingly abundant. In my free time, I enjoy playing and watching soccer (an Arsenal fan, unfortunately), lifting weights, and working on improving my reading and writing skills in a world of AI where original thought is invaluable.

WORK

EXPERIENCE

Grepr

SWE Intern • San Francisco, CA • Summer 2025

  • ●Assisted in building Grepr's new distributed tracing product, handling the query and storage layer for unprocessed traces while maintaining constraints in a distributed systems environment built with Apache Flink, Apache Iceberg, and a React/TypeScript frontend
  • ●Built both the OpenTelemetry log and trace integrations, implementing efficient serialization/deserialization with minimal garbage collection, sustaining 10,000+ records/sec across a Kubernetes-orchestrated AWS infrastructure, ensuring low latency and fault tolerance at scale
  • ●Wrote and executed database migration scripts on production clusters, ensuring schema consistency and safe rollouts
  • ●Built a comprehensive integration test suite in JUnit validating interoperability with Datadog, Splunk, OpenTelemetry, Sumo Logic, and AWS S3

PROJECTS

Distributed File System (DFS)↗

Peer-to-Peer Network • Go

  • ●A decentralized, peer-to-peer distributed file system implementing Content-Addressable Storage (CAS) with AES encryption
  • ●Features automatic file replication, fault tolerance, and modular architecture supporting custom transport protocols and storage backends
  • ●Built with TCP transport layer and SHA-1 cryptographic hashing for secure file operations across network peers

PhishSchool (BostonHacks 2025)↗

FastAPI • Supabase • Vite • Gemini Flash 2.5 • SendGrid

  • ●Interactive platform that helps users learn to spot and avoid phishing with Learn, Detect, and Campaign modules
  • ●AI-assisted detection analyzes .eml emails or any image format to provide a giving users a phishing score, risk level, and key indicators
  • ●Sends custom generate phising emails to users on chosen frequency, keeping track and learning of user failures
  • ●Vite: React/TypeScript + Tailwind (frontend) and FastAPI/Node/Supabase (backend), integrating Google Gemini and Twilio's SendGrid; deployed on Vercel