- Leading AI team at Aila (Riyadh, KSA) developing autonomous AI agents for personalized e-learning
- Fine-tuning LLMs like DevStral for automated bug localization with state-of-the-art performance
- Building GraphRAG systems with multi-agent architectures for enhanced semantic retrieval
- Scaling FAISS databases with distributed HNSW/IVF indices for 1M+ embeddings
- 🏆 Outperformed DeepSeek R1 with DevStral fine-tuning: MAP 0.654, MRR 0.720
- ⚡ 3.56ms query latency on distributed vector databases with 1M+ embeddings
- 📈 +32% improvement in semantic retrieval using Self-Reflection + GraphRAG
- 🔬 Published 4+ research papers in medical AI and computer vision
🔹 Lead AI Engineer @ Aila (Riyadh, KSA) May 2025 - Present
- Leading autonomous AI agent development for personalized e-learning platforms
- Architecting scalable AI systems for educational technology
🔹 Senior ML Engineer @ AiGot (PISA, Italy) Jan 2024 - May 2025
- Fine-tuned DevStral-Small-2505 outperforming DeepSeek R1 across all metrics
- Built distributed graph extraction pipeline processing 3-4 repos daily (1,077+ repos total)
- Implemented GraphRAG-based multi-agent systems improving retrieval by +32%
- Scaled FAISS vector databases achieving 3.56ms query latency on 1M+ embeddings
🔹 AI Research Engineer @ Allied Medical Sciences May 2022 - July 2024
- Developed FFPE (Fast Fourier Positional Encoding) for weakly supervised learning
- Built attention-based neural networks achieving 86% Dice score in segmentation
- Created tools for medical image processing with 1000 samples/2min throughput
|
|
|
|
|
|
|
|
|
🔸 Coding Focus Areas:
- AI/ML Development: PyTorch, TensorFlow, HuggingFace
- LLM Engineering: Fine-tuning, RAG Systems, Multi-Agent AI
- Systems Programming: Rust, C++, Distributed Computing
- Research: Computer Vision, Medical AI, Graph Neural Networks
🔸 Professional Impact:
- 7+ Years Python expertise with enterprise-scale deployments
- 1M+ Embeddings processed in distributed vector databases
- MAP 0.654, MRR 0.720 achieved in DevStral fine-tuning
- 32% Improvement in semantic retrieval performance