Research-driven Data Scientist & ML Engineer with a PhD focus on fault tolerance in high-performance computing. I apply data science, machine learning, and large language models (LLMs) to solve complex problems, from building predictive models and analytics pipelines to designing dependable systems. My work bridges practical data-driven insights with research-level rigor—making complex pipelines feel simple and impactful.
Aug 2025 – Present • Advisor: Prof. Guanpeng Li • Focus: dependable HPC, fault tolerance, GNN, DBN, SDC modeling, LLVM, Fault Injection.
Current
Aug 2024 – 2025 • Advisor: Prof. Guanpeng Li • Focus: dependable HPC, fault tolerance, LLVM-based fault injection & SDC modeling.
Transferred CGPA 4.0
2019 – 2023 • CGPA 3.78 • Dean’s Honor List (2019–2022). Core: ML, DL, AI, Speech, Data Mining, HCI, Security, SE, Automata, NCC.
Honors CGPA 3.78
Developed scalable credit card fraud detection models using ensemble learning, SMOTE, and cloud-based tools.
Built a knowledge-based chatbot with RAG, ChromaDB, and GPT-3.5, achieving ~85% accuracy on domain-specific queries.
Predictive churn modeling with engineered RFM/tenure features and uplift analysis.
Supervised pipelines with SHAP explainability to power targeted campaigns.
Built an interactive geospatial web app by automating data collection to GeoJSON and visualizing it with Flask + Leaflet.
Implemented a MNIST digit classifier using ONNX Runtime Web with WebGL acceleration and WASM fallback.
Multilingual text classifier using classic ML and neural models.
Custom language interpreter with lexer, parser, and runtime in Python.
Teacher-student compression on CIFAR-10 with comparative pipelines.
Cross-platform university portal with multiple student-facing features.
Research-driven mobile UX prototype emphasizing empathy and accessibility.
Classic CV pipeline for celebrity recognition with a lightweight Flask demo.
Misinformation detection combining feature engineering and pattern mining.
Interactive snake with OOP design, collision logic, and difficulty scaling.
ML × HPC · fault tolerance · LLVM/LLFI
Researched fault tolerance at the intersection of compiler analysis and ML, developing LLVM-based tools to study silent data corruption
Supported delivery of the Artificial Intelligence course, mentoring students and enabling stronger outcomes through hands-on guidance.
Contributed to DebloatBench, a unified framework for evaluating container debloaters across diverse design and execution environments, in collaboration with SRI and University of Arizona.
Delivered ML & LLM solutions for global clients while leading training and enablement efforts on modern NLP/LLM practices.
Credit-card fraud detection with Clariba SEIDOR during my time at LUMS (ISPL); built scalable ML pipelines on large, real-world transaction data.
Built automated geospatial data pipelines and interactive visualization tools for constituency-level insights in Pakistan.
Applied ML and analytics to measure branch performance and employee effectiveness while supporting internal recognition programs.
Served as a Peer Advisor at LUMS, mentoring undergraduate students on course planning and academic success strategies.
Teaching Assistant for Computational Problem Solving at LUMS, supporting over 100 students in programming fundamentals.
ETL optimization · SQL · predictive analytics
Draw below and hit Predict.
