AI / ML Engineer
I build and study modern AI systems, with interests in model internals, inference, evaluation, and reliable ML engineering.
I am currently completing a Specialized Master's in Artificial Intelligence at Télécom Paris, after an engineering degree in Computer Science from IMT Nord Europe. I am interested in understanding modern AI systems beyond high-level APIs, from model architecture to inference and deployment.
Recently, I rebuilt GPT-2 from first principles in PyTorch and studied GPU execution through Triton kernel implementation and benchmarking. Previously, I worked on LLM infrastructure and MLOps at Air Liquide.
I am currently looking for a six-month AI/ML engineering or research engineering internship starting January 2027.
PyTorch · Transformers · Model Internals
Reimplemented GPT-2 inference from scratch in PyTorch, including causal self-attention, transformer blocks, weight loading, and autoregressive generation, while validating intermediate representations against a reference implementation.
PyTorch · Computer Vision · PatchCore
Built an unsupervised visual anomaly-detection pipeline for industrial automotive inspection using frozen ResNet features, patch representations, memory-bank retrieval, and nearest-neighbor anomaly scoring.
Reimplementing GPT-2 and validating the architecture against a reference implementation.
What implementing softmax taught me about GPU execution, benchmarking, and shape-dependent performance.
Specialized Master's in Artificial Intelligence
AI / MLOps Intern — LLM infrastructure, AWS Bedrock, MCP, and cloud deployment.
Engineering Degree in Computer Science
I'm always happy to discuss AI engineering, research, open-source projects, or internship opportunities.
lounis.hamroun@gmail.com