I am an AI Engineer & Full Stack Developer, currently in my final year pursuing a BS in Computer Science at Habib University (Batch of 2026). My work bridges the gap between deep learning research, scalable software engineering, and low-level system architecture.

• 🔭 I’m currently working as an AI Engineering Intern at Motive, developing end-to-end testing frameworks for embedded systems and porting code to Linux POSIX environments.
• 🚀 I am the Optimization Team Lead for TurboDiff, building a differentiable fluid dynamics simulator for airfoil shape optimization using JAX and PhiFlow.
• 🏆 ICPC Asia West Finalist (2024-25): Ranked 1st in Pakistan in Round 3 and 2nd in Round 2.
• ☕ When I'm not training models or debugging C++, you can usually find me filming some short film/commercial, hosting an event as the Master of Ceremony, brewing + sipping coffee, watching thriller movies, or reading classical Urdu poetry by Mir, Ghalib and other legends.

• Cybersecurity & Cryptography: Threat modeling and architectural security analysis (Microsoft Threat Modeling Tool), implementing secure cryptographic communication protocols (SHA-256).
• Artificial Intelligence & ML: Natural Language Processing (NLP), Large Language Models (LLMs), Computer Vision, and developing differentiable physics simulations.
• Low-Level Systems & Hardware: Advanced C++ data structures, pipelined processor design (Verilog), and embedded systems testing in Linux/POSIX environments.
• Scientific Computing: Computational Fluid Dynamics (CFD) optimization, Eulerian grids, and high-performance mathematical modeling using JAX.
• Full-Stack & Mobile Engineering: Scalable backend architectures, third-party API integration, and cross-platform mobile application development.

• Languages: Python, C/C++, SQL, JavaScript, Dart
• AI & ML: PyTorch, JAX, HuggingFace Transformers, ModernBERT, LLMs (Gemini, Llama, OpenAI)
• Web & Mobile: Next.js, NodeJS, Flutter, Webflow
• Databases & Systems: PostgreSQL, Docker, Verilog

• Early Detection of Depression (eRisk 2025 Task 2): Developed an asynchronous NLP pipeline utilizing ModernBERT and Llama 3.1 embeddings. Secured 3rd position globally.
• Food Hazard Detection (SemEval 2025 Task 9): Fine-tuned transformers (BERT, RoBERTa) and utilized LLM-based data augmentation to resolve class imbalances. Secured a 4th/5th global rank and co-authored a paper accepted for ACL 2025.
• Recurring Donations System: Architected a scalable backend utilizing NodeJS and PostgreSQL for Green Island Trust, successfully routing transactions through multiple payment gateways including Stripe and PayFast.
• Advanced Data Structures in C++: Implemented and visualized Sleator-Tarjan Dynamic Trees integrated with Dinit's algorithm to optimize network flow calculations.

• Email: [email protected]
• LinkedIn: linkedin.com/in/meesuma5