I'm a Computer Science student at FAST-NUCES focused on building reliable AI systems — from research and evaluation all the way to production.
I work across the full stack of modern AI: data pipelines, model training & evaluation, RAG / agentic architectures, backend services, and deployment. I care equally about rigorous experimentation and shipping systems that actually work under real constraints.
Core strengths
- Multilingual & low-resource NLP (especially Urdu / Roman Urdu)
- Retrieval-Augmented Generation and agentic workflows
- Evaluation design, robustness testing, and shortcut detection
- End-to-end ownership (data → model → API → monitoring)
Open to roles in AI Engineering, ML Engineering, AI Research, Software Engineering, Forward Deployed Engineering, Data Science, and Generative / Agentic AI.
|
Domain Robustness of Multilingual NLP Models • Built & released 33 datasets (288,899 examples) |
When Benchmarks Mislead • Identified critical length confounds in fake news detection |
|
Regime-Aware Multi-Modal Bitcoin Direction Prediction • Proposed RAML (FinBERT + BiLSTM + volatility-conditioned fusion) |
|
| Role | Organization | Focus |
|---|---|---|
| Research Intern | FAST School of Computing | Led independent research on Urdu NLP robustness • Built large multi-domain benchmark • Designed evaluation protocols |
| IT Intern | QuickFuel | Introduced AI/ML into a traditional .NET stack • Built sentiment analysis + semantic search prototypes • Performed system audits |
| AI Automation Intern | Optimus Automate | Built LLM-powered email triage agent and RAG document Q&A system end-to-end |
| Project | What it does | Stack | Link |
|---|---|---|---|
| TaxChronos | Production-style multilingual RAG for Pakistani tax queries (English–Urdu code-switching) | LangChain, Qdrant, Gemini, RAGAS | GitHub |
| Daily Digest | Live AI news aggregator with personalized scheduled delivery (15+ users) | FastAPI, PostgreSQL, Gemini, Docker, JWT | GitHub |
| Urdu Domain-Robustness Benchmark | Open-source 33-dataset dual-script benchmark + evaluation tooling | Transformers, XLM-RoBERTa | GitHub |
| Bilingual WhisperBench | Zero-shot evaluation of Whisper models on code-mixed speech + custom SPER metric | Whisper, FastAPI | GitHub |
| FinSight | Financial risk analysis platform combining anomaly detection, behavioral modeling & NLP | React, FastAPI, Scikit-learn, HMM | GitHub |
- AI Engineer Core Track (LLM Engineering, RAG, QLoRA, Agents) — Udemy
View Certificate - Machine Learning with Python — IBM
View Certificate - Data Analysis and Visualization with Python — Microsoft
View Certificate - Introduction to Generative AI — Google
View Certificate
- Team Lead / Scrum Master on FYP and multiple semester projects
- Media Officer — SOFTEC'25 & SQL Query Master Competition
- ACM Speed Programming Competition
- Delivered AI masterclasses and workshops
- Taught MS Office at Pact College
- Deputy Head of Public Relations — FAST Business Club