AI & Data Science

Alwaleed
Almutairi

I build machine-learning and LLM systems that work in Arabic and English — from bilingual semantic search to computer vision running on a drone.

View projects Download CV
Portrait of Alwaleed Almutairi
Open to Data Science & AI roles

Who I Am

Data Science & Applied AI

I'm a Computer Science graduate from KFUPM. I like problems where the model is only half the work — the other half is the data, the evaluation, and getting it running somewhere real.

Lately that has meant a bilingual hybrid-search platform where the right Arabic article ranks first for 93% of test questions, a YOLOv8 crack detector running on a drone's Raspberry Pi, and RAG apps that answer only from real documents. My internship at Waseel taught me to ship: Angular, Spring Boot, and PostgreSQL in an Agile team.

Arabic questions answered by the top search result
93%
LLM-drafted articles needing only light edits
85%
crack types classified on-device
8
Download CV
Selected Work

Projects

Resolvix
In progressJun 2026 – Present

Resolvix

A bilingual knowledge platform for support teams — hybrid search and a Q&A assistant that only quotes approved articles.

Arabic questions where the correct article ranks first
93%
English drafts needing only light edits in testing
85%
  • Hybrid search that combines keyword matching with multilingual embeddings, plus a Q&A assistant that only quotes approved articles.
  • Evaluated embedding models on hand-written Arabic and English queries before choosing one.
  • Local LLM pipeline that redacts personal data, clusters similar tickets, and drafts new articles.
  • Approval workflows, role-based access, and a Freshdesk integration.
  • React
  • Node.js
  • PostgreSQL
  • Multilingual embeddings
  • Ollama
Website Resolvix (opens in a new tab)
Illustration
CompletedJan 2026 – May 2026

Drone-Based Structural Crack Detection

AI Model Developer · Senior Design Project, KFUPM

A drone inspection system that detects and classifies eight types of structural cracks — with the model running on board.

crack types: shear, flexural, torsion, settlement…
8
unlabeled images pseudo-labeled
~3K
  • Merged two datasets into one 8-class crack taxonomy — remapped class IDs, rebalanced skewed classes, and kept clean train/valid/test splits.
  • Warm-started the 8-class YOLOv8 model from a 3-class model for faster, more stable training.
  • Built a pseudo-labeling pipeline to scale annotation across ~3,000 unlabeled images.
  • Deployed inference on a Raspberry Pi 5 with a RealSense depth camera, and wrote the Python mission scripts for autonomous inspection.
  • Python
  • YOLOv8
  • PyTorch
  • Raspberry Pi 5
  • Intel RealSense D435
Source Drone-Based Structural Crack Detection (opens in a new tab)
TalkToPDF
LiveJan 2026 – Feb 2026 · upgraded Oct 2026

TalkToPDF

Chat with your PDFs and get answers that cite the exact page — parsing, indexing and search all run in your browser.

recall@3 on its retrieval benchmark, up from 79% with TF-IDF
93%
English benchmark questions answered from the right passage
25/25
  • Hybrid search in the browser: BM25 keyword search plus multilingual embeddings in a Web Worker, fused with Reciprocal Rank Fusion.
  • Clickable citations — every [n] in an answer opens the exact passage and page the model saw.
  • Works in Arabic: right-to-left PDF text is rebuilt correctly, and you can ask in Arabic about an English document.
  • Your files never leave the device; a Vercel function answers with Groq and falls back across models when one is busy or retired.
  • React
  • Vite
  • BM25
  • multilingual-e5-small
  • Groq API
Live demo TalkToPDF (opens in a new tab) Source TalkToPDF (opens in a new tab)

More projects

My Journey

Experience & Education

Experience

  1. Jun 2025 – Aug 2025

    Software Engineering Intern

    Waseel

    • Built an internal portal with Angular, Spring Boot, and PostgreSQL to streamline ticket management.
    • Integrated the HubSpot API to sync tickets automatically, eliminating manual platform switching.
    • Gathered client requirements and iterated on direct feedback in an Agile/Scrum team.
  2. Jan 2024 – Jan 2025

    Leader of Public Relations

    KFUPM Computer Club

    Organized tech workshops and events with 200+ attendees, then led public relations and outreach campaigns.

Education

  1. 2021 – 2026

    B.Sc. Computer Science

    King Fahd University of Petroleum and Minerals (KFUPM)

  2. 70 hours

    Artificial Intelligence Program

    KAUST Academy

    Machine learning foundations, supervised and unsupervised learning, and neural networks.

Skills

AI & Data
  • Machine & Deep Learning
  • LLMs & RAG
  • Semantic Search
  • Model Evaluation
  • Ollama (Local LLMs)
  • Data Analysis
Engineering
  • Python
  • Java
  • SQL
  • PostgreSQL
  • React, Node.js, Express
  • Angular, Spring Boot
  • Git, Docker
Languages
  • Arabic
  • English

Competitions

Participant

Consulting Championship 2025

Aramco · Roland Berger · Deloitte · EY

Ask Me Anything

AI Assistant

Answers come from a knowledge base about my work — retrieved with hybrid search (pgvector + full-text) and written by an LLM.

Hi! I'm Sanad, Alwaleed's AI assistant. Ask me about his projects, skills, or experience — or how to reach him.

Get In Touch

Contact Me