Uzafir A. Rafaq

Mathematician and Computer Scientist

Research Profile

Electrical and software engineer pursuing research in Physics Informed Machine Learning and Human-Computer Interaction. Research deals with statistical mechanics, probabilistic modeling, dynamical systems, and human participant experiments. Experienced in turning models into deployed, reproducible software systems.

Statistical mechanicsArtifical IntelligenceHuman-Computer InteractionHigh Performance Computing

Education

  • MSc Software Engineering2025 to 2026
    Heriot-Watt UniversityDubai, United Arab Emirates
    • Dissertation: Thermodynamic Human-Computer Interaction.
    • Developed a target prediction model using Hamiltonian, Lagrangian and statistical mechanics; implemented and evaluated the model across desktop and mobile interaction.
    • Derived the 70 year old empirical Fitts' Law and Schmidt's Law from thermodynamics.
    CodeDemo
  • BSc Electrical Engineering2018 to 2022
    Institute of Space TechnologyIslamabad, Pakistan
    • Gold Medal for Best Final Year Project.
    • Final project: sign language glove using machine learning and the Google Translate API to convert hand gestures into speech in multiple languages.
    CodeVideo

Awards and Honours

  • Gold Medal for Best Final Year Project2022
    Institute of Space TechnologyIslamabad, Pakistan

Invited Talks

  • Thermodynamic Artificial Intelligence13 October 2026
    National Autonomous University of MexicoOnline

    Invited lecture for the Complexity Theory seminar on the thermodynamic aspects of artificial intelligence.

Skills

Languages

Python
JavaScript
TypeScript
SQL

Frameworks

React
Next.js
Django
Express
NestJS
Docker
WebRTC

Databases

PostgreSQL
MongoDB
Redis

Cloud

Google Cloud Platform
AWS
Azure
DigitalOcean

Projects

  • Intent Link2026
    intentlink.dev
    • A user target prediction library from MSc research turned into an open-source library for React and Next.js.
    • The world's most accurate and efficient prefetching library with an accuracy of 98% and a fetch-to-click ratio of 1.37.
    npm packageDemo
  • Beep2024
    beep.gov.pk
    • Implemented WebRTC with a MediaSoup signaling server to create audio and video calling for the military.
    • Enforced strict cybersecurity best practices to align with defense industry compliance standards.
    • Implemented microservices and microfrontends in Next.js and Express.js.

Experience

  • Ordion.ioRemote, United States
    Software EngineerDecember 2024 to August 2025
    • Developed caching and prefetching strategies using Next.js and Redis application workloads.
    • Contributed to the conversion of a monolithic application into independently deployable services.
    • Built containerized deployment pipelines for Google Cloud and implemented load balancing, content delivery and automated SSL certificate management.
  • Linter LabsIslamabad, Pakistan
    Software EngineerOctober 2023 to December 2024
    • Led API development using Django, NestJS, and ExpressJS.
    • Implemented real time messaging and group audio and video communication using Redis, Socket.io, WebRTC, and a media server.
    • Deployed containerized applications through Azure Pipelines to Google Cloud and deployed services on AWS infrastructure.
  • ML Sense SolutionsIslamabad, Pakistan
    Software EngineerMarch 2022 to October 2023
    • Developed web interfaces and backend services using React, Next.js, Django, and ExpressJS.
    • Implemented authentication, real time communication, and cloud data storage using Firestore, Cloud SQL, and BigQuery.
    • Automated container deployment to Google Cloud and conducted cloud cost analysis.

Publications

  1. [1]Uzafir Ahmad Rafaq, Muaz Hassan, and Ali Muzaffar. 2026. Thermodynamic Human-Computer Interaction. arXiv:2608.07123 [cs.HC]. Under review at ACM CHI.