Work Experience

Research Software Engineer · University at Buffalo (SUNY)

• Building a transaction reordering layer in Go that lowers tail latency and aborts of existing strictly serializable concurrency control without changing its code: up to 75% lower tail latency and 85% fewer aborts across 2PL, OCC and NCC on CloudLab; in preparation for USENIX OSDI 2027.
• Prototyped primary-backup, chain replication, and distributed two-phase locking in Go on CloudLab (NSF-supported).
• Building a transaction protocol in Go and Java for microservices that each keep their own, possibly different, database, aiming for serializable, all-or-nothing transactions without requiring XA support and without the row locks Apache Seata’s AT and XA modes hold until the global commit; in preparation for USENIX OSDI 2027.

Software Engineer · Linq.io (remote)

• Engineered a playground and diagnostics platform for an asset-management system: workflow simulation, feature validation, and API testing, with observability through structured logging, distributed tracing, fault isolation, and real-time monitoring.
• Shipped an on-demand media-generation API (FastAPI, MongoDB) that cut page load by 60% by generating asset thumbnails as users browsed 10K+ inventories.
• Integrated OAuth 2.0 SSO for enterprise accounts so customers could sign in with existing corporate credentials.
• On call twice a week for production pipelines and client issues, tracked in Jira; weekly PR reviews with senior engineers.

System Engineer · Its IT Group (contract, remote)

• Redesigned the Node.js / MongoDB backend of an AR app for VTuber/karaoke 3D avatars with WebSockets for real-time comments, notifications, and likes across 5K+ daily active users.
• Implemented a 3D model pipeline for LiDAR scans with a USDZ → GLB converter for Unity integration, cutting manual measurement time by 80%.

Software Engineer · Interactive Media Lab, LUMS

• Built Awaaz-e-Sehat, a HIPAA-aligned e-health platform on Flask and AWS Lambda, improving API latency from 5s → 500ms for 20K+ daily active users.
• Productionized Whisper + GPT-4 transcription into structured clinical notes (95% accuracy, 1K+ recordings/day).
• Ran a serverless ingest and search stack (Lambda, DynamoDB, S3) with automatic scaling; Redis caching reduced API latency by ~30%.

Research Software Engineer · Network and Systems Group, LUMS

• Released a public dataset covering 17.6K videos and 46.6K ads across 8 countries with per-video buffer telemetry, the first large-scale YouTube corpus with that coverage.
• Measured how YouTube main-video bitrate drives buffer loss across resolutions (720p ~3× higher latent buffer loss than 360p; 10.1 MB vs 3.4 MB).
• Built the distributed collection and analysis toolchain (Selenium/PyTube) behind the corpus and released it publicly alongside the dataset.
• Published at The ACM Web Conference 2024 (WWW ’24): doi:10.1145/3589334.3645496.

Teaching Assistant · University at Buffalo

• Modern Networking Concepts (CSE 489/589)

Teaching Assistant · Lahore University of Management Sciences

• CS200: Object Oriented Design in C++ (Spring ’23, Fall ’23)
• CS582: Distributed Systems (Fall ’23)

PhD, Computer Science · University at Buffalo (SUNY)

• Available for Summer 2027 software engineering / research internships, returning to the PhD afterwards.

BS, Computer Science · Lahore University of Management Sciences (LUMS)

Participant · Y Combinator AI Startup School

• Selected from a pool of 30,000 applications for YC’s AI Startup School 2026 at Chase Center, a program for builders working on AI products and research.
• Joined office hours and 1:1 conversations with founders and operators about what we’re building; networked with peers across AI systems, products, and research.

Backend / frontend coursework projects & short roles · LUMS

P2P file sharing with DHTs, speech-therapy MERN app (Guftaar), and a short course-website engagement for an Internet Architecture workshop.

Machine Learning Developer (freelance) · Fiverr

Dataset cleaning, model comparison, and end-to-end ML pipelines for client projects.