Summary
Overview
Work History
Education
Skills
Certification
Languages
Timeline
Generic

TAIF NONO

Riyadh, Saudi Arabia

Summary

AI Specialist | ASR, Computer Vision, Transformers, Experienced in Deep Learning, ASR, Computer Vision, and Transformer-based models. Skilled in building AI solutions for automation and pattern recognition. Passionate about impactful, real-world applications in collaborative settings.

Overview

3
3
years of professional experience
1
1
Certification

Work History

AI & ML Engineer

Confidential Government
07.2023 - Current
  • Built end-to-end Optical Character Recognition (OCR) systems to extract text from scanned documents and images, specializing in the Arabic language.
  • Develop and Fine-tune recognition models for Arabic OCR with diverse architectures, including CRNN (ResNet + LSTM + CTC), Transformer-based CTC, Vision Transformer with a GPT-style decoder, and vision-language models (Qwen2.5-VL).
  • Trained layout and detection models using SegFormer to accurately segment and detect text regions in complex Arabic documents.
  • Built a complete pipeline for synthetic Arabic OCR data generation, creating over 10M documents with diverse layout styles and applying an advanced augmentation pipeline to simulate real-world document variability, and using open source tool like SynthTiger.
  • Benchmarked Arabic OCR models components (layout, detection, and recognition) using both automated metrics (Pixel Accuracy, IoU for layout/detection; WER and CER for recognition) and manual evaluations to ensure robustness and accuracy.
  • Trained and fine-tuned Automatic Speech Recognition (ASR) systems to transcribe spoken Arabic into text, leveraging advanced architectures such as Squeezeformer and Conformer with speaker diarization capabilities using NVIDIA NeMo, enabling separation and identification of different speakers in Arabic audio.
  • Evaluated ASR performance by benchmarking models with Word Error Rate (WER) and Character Error Rate (CER), ensuring accuracy and reliability for Arabic speech transcription.
  • Built LangChain pipelines integrating LLMs with few-shot learning to process structured data and synthetically generate ITN (Inverse Text Normalization) Arabic data, converting spoken-style text into standard written form.
  • Built automated training and evaluation pipelines for AI models on distributed multi-node, multi-GPU systems.

Senior Project

Jeddah University
11.2022 - 02.2023
  • Developed a user-friendly interface to calculate companies’ social listening metrics and sponsorship effectiveness by detecting logos in visual content, enabling data-driven marketing decisions.
  • Built a two-stage detection and feature extraction system using Faster R-CNN for logo detection and VGG16/ResNet for extracting features from detected logos and query images, with similarity measured via cosine similarity.
  • Leveraged large-scale datasets such as “Logos in the Wild” to train the models for better generalization across diverse logos, improving detection performance and robustness.

Intern

Saudi Federation for Cyber Security and Programming
06.2022 - 08.2022
  • Collected, cleaned, and explored datasets to support a custom Logo Detection project.
  • Reviewed research papers and explored state-of-the-art detection models.
  • Prepared and converted annotated data from XML to YOLO and COCO formats for training.
  • Implemented and optimized Detectron2 models on custom datasets through multiple experiments.
  • Developed and evaluated Transformer-based DETR models for logo detection.
  • Dockerized the project and deployed models via a Flask API for production use.

Education

Bachelor of Science - Artificial intelligence

University of Jeddah
Jeddah, Saudi Arabia
01.2023

Skills

  • Python
  • Bash
  • PyTorch
  • Flask
  • Docker
  • Machine Learning
  • Deep Learning
  • MLOps
  • Model Deployment
  • LangChain
  • Data Generation
  • Streamlit
  • MinIO

Certification

  • NVIDIA-Certified Associate: Generative AI LLMs NVIDIA - 2025
  • NVIDIA-Certified Associate: Gen AI Multimodal NVIDIA - 2025
  • Introduction to Transformer-Based Natural Language Processing NVIDIA
  • Building RAG Agents with LLM NVIDIA
  • Fundamentals of Accelerated Data Science NVIDIA

Languages

Arabic
Native or Bilingual
English
Full Professional

Timeline

AI & ML Engineer

Confidential Government
07.2023 - Current

Senior Project

Jeddah University
11.2022 - 02.2023

Intern

Saudi Federation for Cyber Security and Programming
06.2022 - 08.2022

Bachelor of Science - Artificial intelligence

University of Jeddah

Similar Profiles

CREATE PROFILE
TAIF NONO