- 👋 Hi, I’m @ahmedheakl
- 🎓 PhD student in Computer Vision at MBZUAI, advised by Prof. Salman Khan and Prof. Fahad Khan
- 👀 I’m interested in Computer Vision 🧠, NLP 🌐, Software Engineering 📱
- 🌱 I’m currently learning self-evolving VLMs 🎼.
- 👨💼 I'm currently a Research Scientist Intern at 🇰🇷 LG AI Research 🌍 .
- 👯 I’m always looking to collaborate with other researchers in the field of Deep Learning
- 📰 Checkout my latest publish paper: Dr.LLM: Dynamic Layer Routing in LLMs
- 📰 Checkout my latest Python packages: DroneVis 👾.
I am currently working on developing efficient-accurate LLMs Dr.LLM, OCR models DocAtlas, Multilingual Models AIN, reinforcement learning for multiple vehicles optimizations SVRPBench, and GPU-to-GPU translation using LLMs CASS.
| Paper | Venue |
|---|---|
| CEPO: Contrastive Evidence Policy Optimization | Preprint 2026 |
| Dr.LLM: Dynamic Layer Routing in LLMs | ICLR 2026 |
| SAHM: Arabic Financial & Shari'ah Reasoning | ACL 2026 (Oral) |
| CASS: Nvidia to AMD Transpilation | ACL 2026 |
| KITAB-Bench: Arabic OCR & Document Understanding | ACL 2025 |
| SVRPBench: Stochastic Vehicle Routing | NeurIPS 2025 |
| GG: CISC-to-RISC Transpilation | EMNLP 2025 |
| CAMEL-Bench: Arabic LMM Benchmark | NAACL 2025 |
See full list on Google Scholar.
| Project | What it does |
|---|---|
| CEPO | Token-level credit assignment for RLVR |
| DroneVis | Computer vision pipeline for drones |
| AIN | Arabic/English large multimodal model |
| Mobile-O | On-device unified VLM + image generation |