About
I’m Ahsan Umar, an AI/ML Engineer & Researcher passionate about Generative AI, Large Language Models (LLMs), Retrieval-Augmented Generation (RAG) systems, and Computer Vision. My work bridges research and practical deployment, making advanced AI systems more accessible, efficient, and impactful.
Over the years, I’ve gained expertise in:
- LLMs & Generative AI – fine-tuning transformer-based models (QLoRA, PEFT), building training pipelines, and optimizing inference efficiency.
- RAG Systems – designing and deploying production-ready pipelines with LangChain, FAISS, Pinecone, and ChromaDB, reducing latency and improving retrieval accuracy.
- Healthcare AI & Computer Vision – building CNN and Transformer-based models for medical imaging tasks with explainability (Grad-CAM).
- MLOps & Deployment – scaling AI applications on AWS (SageMaker, Lambda, EC2) and GCP (Vertex AI, Cloud Run) with CI/CD, monitoring, and serverless deployment.
- Open Source Contributions – publishing tools and research such as QuantLLM (efficient LLM fine-tuning with quantization), DiffusionLM, and experimental transformer architectures.
I actively share educational resources, open-source implementations, and research-driven projects to lower barriers for students, developers, and researchers entering AI.
My mission is to:
- Democratize AI education by creating transparent, step-by-step learning resources.
- Advance AI research in low-resource NLP, multimodal systems, and healthcare.
- Build sustainable, open ecosystems where anyone can learn, contribute, and deploy AI responsibly.
📚 Education: BS in Artificial Intelligence (Islamia College University, Peshawar, 2023–2027).
💡 Certifications: Deep Learning (Coursera), Generative AI (AWS), NLP with Transformers (Hugging Face), Google TensorFlow Developer, and more.
💡 Certifications: Deep Learning (Coursera), Generative AI (AWS), NLP with Transformers (Hugging Face), Google TensorFlow Developer, and more.