Sanjay

Carnegie Mellon University · Pittsburgh

Applied AI, backend platforms, and distributed systems.

I've designed, shipped, and operated production software since 2022. That includes an AI system that conducts real-time voice and video interviews with job candidates, backend services supporting 10,000+ concurrent clients, database infrastructure that cut cloud costs by 80%, and deployment automation that reduced release time from 30 minutes to 8.

Explore selected work Start a conversation

01

Selected work

Professional, independent, and graduate projects that show the systems I have built.

Backend infrastructure · Professional

Real-time AI platform

Live product

Built real-time messaging for 10,000+ concurrent WebSocket connections, authentication across 20+ AI agents, high-throughput SQL, a lower-cost database platform, and automated deployment workflows. The product reached 1.5M+ unique wallets.

10,000+ concurrent clients · 66,000 ops/sec · 80% lower cloud cost · releases 30 → 8 min

  • TypeScript
  • PostgreSQL
  • AWS
  • WebSockets
  • GitHub Actions

Applied AI · Professional

Real-time AI interviewing system

Built a voice-and-vision AI interviewer with cascade model routing targeting sub-second responses. Added a test harness that ran realistic candidate scenarios, recorded tool calls and outputs, and verified interview-progression rules. Kept the model layer compatible with open-source and on-premises deployment.

Internal pilot · sub-second response target · open-source and on-premises compatible

  • Python
  • Pipecat
  • Temporal
  • Evaluation

Distributed systems · Carnegie Mellon

Distributed migration engine

Built a three-node Raft coordinator in Go with leader election, replicated state, dependency-aware scheduling, idempotent replay, and progress streaming. Tested crash, partition, dynamic-join, and leader-failover scenarios.

Leader failover · partitions · dynamic joins

  • Go
  • Raft
  • RPC
  • Concurrency

Independent production system

Valorina

View source

Built and operated a Python service for Valorant store tracking. Scaled the system with sharding and dynamic EC2 capacity to remain responsive through high-traffic periods.

875,000+ users · 15,000 servers · independently operated

  • Python
  • MongoDB
  • AWS EC2
  • Sharding

Fintech · Four-person team

Scotty Market

Team repository

On a four-person team, built mobile and backend features for an AI-assisted personal-finance app, including dashboards, budget projections, recurring-charge detection, and agent-backed guidance.

1st Place, Visa Track · 1,000+ participants · 50+ universities · four-person team

  • React Native
  • TypeScript
  • Express
  • SQLite

More work

Tartan Tickets Built authorization and prompt-injection safeguards, then refactored a monolithic API gateway into five domain routers, removing 578 lines through an expand-contract rollout monitored in Grafana.
Cloud-native bookstore Five microservices with Kafka workflows, circuit breaking, and Kubernetes delivery.

02

Experience

Roles, responsibilities, and results since 2022.

Neuroscale AI Engineering Intern AI interviewing system and workflow automation May 2026 – Jul 2026

Sterling, VA

  • Translated competitive research into product requirements and system design for an AI system that conducts real-time voice and video interviews; shipped an internal pilot tested against realistic candidate scenarios.
  • Engineered a Pipecat voice-and-vision agent with model routing, an evaluation harness, tool-use evidence collection, and controls that gated interview progression.
  • Automated tenant branding, custom domains, and candidate email delivery with Temporal workflows integrated into existing platform services.
Capx AI Founding Software Engineer Backend platform, data, and delivery infrastructure · Previously Software Engineer Intern Aug 2023 – Aug 2025

Dubai, UAE | Hybrid

  • Engineered real-time messaging with AWS WebSocket API, Node.js, and Supabase for 10,000+ concurrent clients using batched delivery and session tracking.
  • Led backend development for an AI-agent marketplace on an Arbitrum Orbit network that reached 1.5M+ wallets.
  • Built OAuth/JWT-based authentication for interactions across 20+ AI agents.
  • Migrated Firestore workloads to a self-hosted relational database, reducing cloud costs by 80%, and designed SQL logic supporting 66,000 ops/sec peak load.
  • Built a GitHub Actions deployment pipeline with Amazon ECR and App Runner, reducing release time from roughly 30 minutes to 8 minutes.
Flowace AI Software Engineer Intern Linux desktop engineering and developer integrations Jul 2022 – Jul 2023

Dallas, TX | Remote

  • Rewrote legacy C modules for an Ubuntu desktop application to support Wayland compositors and shipped the work through beta and production testing.
  • Built a Chrome extension for Asana time logging and OAuth authorization-code integrations using Node.js and LoopBack.

03

Background

Education, teaching, and leadership.

Before graduate school, I worked on high-throughput financial and decentralized systems serving large transaction workloads. I also enjoy mentoring engineers and discussing the trade-offs behind reliable software.

I am completing Carnegie Mellon’s MSE in Scalable Systems. I am particularly interested in backend and platform engineering, applied AI, developer infrastructure, distributed systems, and infrastructure-heavy full-stack roles.

Education

Carnegie Mellon University

Master of Software Engineering · Scalable Systems

Vellore Institute of Technology

Bachelor of Technology in Computer Science

Teaching & community

Incoming Teaching Assistant · 17-695 Design Patterns

Carnegie Mellon University

Treasurer

Carnegie Mellon University · MSE Cohort

Chairperson

Computer Society of India · VIT