Anas Khatiri

Designed and deployed enterprise-grade AI solutions using Generative AI, LLMs, RAG, and AI agents. Built a multimodal document intelligence platform processing PDFs, OCR-scanned documents, HTML, and multimedia content, combining semantic search, vector databases, and agentic workflows for grounded knowledge retrieval. Developed evaluation frameworks and production APIs with FastAPI, Docker, Kubernetes, and AWS, supporting reliable and scalable AI solutions for complex technical documentation.

Built a deep learning-based computer vision system for automated surface crack detection in industrial inspection. Developed image preprocessing, data augmentation, transfer learning, fine-tuning, and model evaluation pipelines using Python, OpenCV, and TensorFlow/PyTorch. Applied knowledge distillation to balance detection performance and inference efficiency, and deployed the model through production APIs using FastAPI/Flask, Docker, and Kubernetes.

Developed AI algorithms for autonomous multi-drone coordination, focusing on task allocation, path planning, and decentralized decision-making. Implemented consensus-based task allocation methods (CBBA/CBAA), reinforcement learning, and optimization techniques in Python and C++ to support autonomous mission planning and coordination. Evaluated multi-UAV scenarios in simulation environments, including Unity 3D.

Designed and implemented trajectory optimization algorithms for autonomous aerial and maritime vehicles, integrating motion constraints, obstacle avoidance, and mission objectives. Applied reinforcement learning and optimization techniques in Python and C++ to generate and evaluate feasible trajectories. Developed simulation-based evaluation pipelines to assess navigation performance and improve trajectory safety, efficiency, and robustness.