خالد آل دوسـري
AI Engineer | NCA-GENL Tuwaiq Academy Alumnus
NVIDIA-Certified Associate: Generative AI LLMs (NCA-GENL), Senior computer science student and Tuwaiq Academy alumnus focusing on AI Engineering and Data Science. Experienced in building end-to-end solutions built with python, data science, machine learning, deep learning, MLOps, LLMs, RAG, generative AI, and agentic systems. Passed the Building and Developing AI Models Bootcamp by Tuwaiq Academy, and completed several courses such as Building transformer-based NLP applications course by Tuwaiq Academy and NVIDIA, as well as other courses and certifications. All of which enabled me to have the skills necessary for building scalable, production-ready systems that turn raw data into actionable insights and real-world impact.
Building and Developing AI Models Bootcamp
June 2026 – September 2026
Bachelor's degree, Computer Science
August 2022 – present
CCIS Student Advisory Council
September 2025 – September 2026
THA by Armonia
January 2026 – May 2026
Raqeeb detects illegal items in X-ray baggage scans, executing an automated incident report. A fine-tuned YOLOv8-OBB model (mAP50 0.919) flags prohibited items, a human in the loop verifies, then a voice agent writes the incident report and calls the responsible authority automatically and independently.
Developed an end-to-end agentic automated machine learning pipeline leveraging Python, LangChain, FastAPI, Next.js, Modal, sandboxing, React and TypeScript; built a full MLOps stack enabling users to upload raw datasets, train custom models, and deploy production-ready cloud APIs via Modal.
Adapted the DreamDiffusion latent diffusion pipeline from 128-channel research-grade EEG to a 20-channel 10-20 layout, an 84% cut in sensor density; re-trained a ViT-Large encoder via Masked Signal Modeling on 120K MOABB signals plus CLIP alignment on 12K ImageNet-EEG pairs, benchmarked on CLIP-Sim against the 0.733 baseline.
Currently designing an agentic recommendation system with python, Node.js, MongoDB, JavaScript, HTML and CSS that parses PDF/text academic transcripts and maps historical grade distributions, therefore increasing recommendation alignment based on user-specific academic history.
Developed an AI-powered tourism analytics platform for the PwC Empowerthon, securing 4th place out of 11 teams. Integrated a bilingual analytics dashboard and ROI calculator using React, TypeScript, Vite, Supabase and JavaScript, aiding Saudi destination owners turn visitor data into actionable insights.
Built a Deep Learning model from scratch classifing the CIFAR-10 dataset by benchmarking four deep learning models with fully connected networks only and no convolution reaching 54.08% accuracy and 53.99% Weighted F1. Measuring exactly how that constraint costs by fine-tuning a MobileNetV2 CNN on the same split reaching 91.24% accuracy and 91.20% Weighted F1.
