I'm Umer — I work with computers, work with models and build AI systems. 🤖💻
I am a Post-Doctoral Researcher / Ingénieur de Recherche specializing in Large Language Models (LLMs), Natural Language Processing (NLP), and Multimodal AI. 🧠
I hold a Ph.D. in Computer Science focused on NLP and transformer-based architectures, with over five years of experience developing, fine-tuning, evaluating, and deploying AI systems using Python and PyTorch. 🔬
My expertise spans:
- 🤖 LLM fine-tuning, inference, merging & optimization
- 🎯 PEFT / LoRA / QLoRA / Quantization
- 🌍 Multilingual & multi-task learning
- 👁️ Multimodal & Vision-Language Models
- 🚀 Large-scale model inference
- 🏗️ End-to-end ML pipelines & AI services
- 🖥️ High-Performance Computing (HPC)
I build AI systems and ML pipelines using technologies such as Hugging Face, Unsloth, LLaMA-Factory, vLLM, FastAPI, Docker, MLflow, and SLURM/HPC infrastructures. 🚀
LLMs · NLP · Multimodal AI · VLMs · Hugging Face · Python · PyTorch · transformers
LLaMA-Factory · Unsloth · PEFT/LoRA · QLoRA · Fine-tuning · Quantization · Model Merging · vLLM · Inference Optimization · Multilingual & Multi-Task Modeling
FastAPI · Docker · MLflow · SLURM/HPC · GPU Computing · Model Serving · ML Pipelines · Git · GitHub Actions
TransformerLens · Captum · Model Evaluation · Interpretability · Activation Analysis
NumPy · Pandas · SQL · Matplotlib · Scikit-learn