About
I am a Computer Science student at FAST National University of Computer and Emerging Sciences (FAST-NUCES), Lahore. My work sits at the intersection of machine learning evaluation, language models, and applied AI systems.
My research focuses on evaluating AI systems in domain-specific contexts and exploring efficient methods for uncertainty estimation in large language models. I am passionate about building robust, real-world AI applications, ranging from autonomous agents to computer vision systems.
Alongside my research, I am actively involved in teaching and mentoring, helping students across the globe build strong foundations in programming and algorithms.
Publications
Evaluating whether lightweight confidence estimates stay reliable across changing domains using controlled train/test comparisons; studying transfer loss and limited-data adaptation with held-out testing rather than relying only on in-domain results.
Selected preprint is available below.
Experience
- Co-designed a reproducible benchmark of four Urdu TTS systems across four speech domains and 960 reference-synthetic audio pairs, combining acoustic, speaker-similarity, and human perceptual evaluation.
- Built evaluation software for MCD, F0 RMSE, speaker embeddings, MUSHRA-style tests, and ABX analysis; found strong domain sensitivity and rank disagreement between objective and perceptual measures.
- Contributed to study design, evaluation software, experimental analysis, and manuscript writing.
- Continuing as RA on AI Reliability Under Distribution Shift: evaluating whether lightweight confidence estimates stay reliable across changing domains using controlled train/test comparisons, and studying transfer loss and limited-data adaptation with held-out testing.
- Working with a research team on agents that continue acting while learning from feedback, connecting reliability detection with adaptation under changing environments.
- Supported 100+ Data Structures and Algorithms students by preparing and grading assignments, conducting vivas, reporting results to instructors, and helping during problem-solving sessions.
- Led 24+ live Python sessions over 8 weeks for students from 15+ countries, using live coding and small-group exercises to teach introductory programming.
- Taught introductory Python, machine learning, and image processing to 20+ students from underserved communities during a one-month training program.
- Reviewed 50+ AI/full-stack projects and provided technical feedback.
- Promoted three global AI hackathons and ranked among the top ambassadors for outreach.
- Coordinated accommodation for 500+ participants during the three-day SOFTEC festival, Pakistan's largest international IT event.
Projects
Selected research and technical projects.
Certificates & Credentials
MIT × Hack-Nation Global AI Hackathon
1st Place · 2,000+ teams from 60+ countries
2026
Ranked 1st among 2,000+ international teams. Built Healthcare Infrastructure Intelligence – India, a 5-agent LangGraph system over 10,000+ facilities that separates medical deserts from data deserts.
Curriculum Vitae
My detailed academic CV.
Download CVContact
Get in Touch
I welcome messages about research collaborations, graduate opportunities, or general inquiries. I typically respond within a few business days.