Computer Vision & OCR
Object detection, image preprocessing, document understanding, OCR integration, MRZ validation, and structured extraction.
Applied AI · Research · Engineering
Mohamad Ali Al Saidi
AI & Machine Learning Engineer · PhD Candidate in AI & Robotics
I build practical machine-learning systems across computer vision and OCR,LLM and multilingual NLP, efficient deep learning, and edge AI—from data preparation and model training to desktop, local, and embedded deployment.
Computer Vision & OCR
LLMs & Multilingual NLP
Model Optimization
TinyML & Edge AI
Profile
Engineering AI from experiments to usable systems
A software-engineering foundation combined with research experience in applied machine learning.
I am an AI and Machine Learning Engineer and PhD candidate in Artificial Intelligence and Robotics. My recent work focuses on computer vision and OCR, LLM, multilingual transformer fine-tuning, model compression, and TinyML.
I work across the full project lifecycle: dataset preparation, model adaptation, training, evaluation, optimization, application integration, and deployment to desktop, local, or embedded environments. My primary tools include Python, Java, PyTorch, TensorFlow, SQL, Hugging Face Transformers, OpenCV, and TensorFlow Lite.
Specialization
Core focus areas
NLP, LLMs & VLMs
Transformer fine-tuning, local inference, translation, tokenization, RAG, multimodal embeddings, and vision-language models.
Efficient Deep Learning
Knowledge distillation, unstructured pruning, quantization, sparsity analysis, and device-specific benchmarking.
TinyML & Edge AI
Sensor-data pipelines, compact neural networks, TensorFlow Lite conversion, quantization, and microcontroller deployment.
Professional Background
Experience
Airline System Developer
Sky Gift
- Developed and maintained a modular airline reservation and operations platform supporting passenger records, bookings, ticketing workflows, and internal administration.
- Built desktop interfaces and operational dashboards using Java, SQL, and related technologies.
- Integrated a passport field-detection and OCR workflow combining computer vision, OCR, MRZ validation, and structured data extraction.
- Developed Python and SQL utilities for data processing, validation, reporting, and information exchange.
- Supported Windows servers, databases, network infrastructure, backups, access control, and operational troubleshooting.
System Administrator Intern
ICT Department, Ferdowsi University of Mashhad
- Installed, configured, and maintained Windows and Linux workstations, servers, and network equipment.
- Assisted with operating-system updates, security patching, account management, backup procedures, and technical troubleshooting.
- Automated recurring administration and maintenance tasks using Bash and PowerShell scripts.
Knowledge Sharing
Teaching & Training
University-level machine learning instruction and practical AI training for educators.
Machine Learning Course Co-Instructor
Ferdowsi University of Mashhad
- Shared semester-long teaching responsibilities with the supervising professor.
- Delivered machine-learning lectures and practical material covering model training, evaluation, and applied problem solving.
AI in Education Workshop Instructor
Hamrah Aval Academy
- Delivered a two-day live workshop series for teachers in two Iraqi cities on practical and responsible AI use in education.
- Covered prompt engineering, teaching-material creation, assessment support, exam preparation, and AI-assisted lesson planning.
Selected Work
Featured AI/ML projects
Four projects selected for technical depth, measurable outcomes, and end-to-end implementation.
PROJECT 01
Passport Field Detection & OCR
A private passport-processing pipeline combining YOLOv8 field detection, OCR, MRZ validation, structured output, and a desktop interface.
92.76%
mAP@0.5
94.26%
Detection recall
604
Passport images
7
Detected fields
PROJECT 02
Fine-Tuned Local Multilingual Translator
An offline Persian–Arabic–English translation application using fine-tuned multilingual transformer weights trained on a private dataset.
3
Supported languages
Local
Offline inference
Private
Training dataset
GUI
Desktop application
PROJECT 03
Model Pruning & Knowledge Distillation
An experimental VGG19/CIFAR-100 compression pipeline comparing baseline training, iterative pruning, and dense versus smaller distilled students.
39.47%
Teacher sparsity
59.3%
Size reduction
16.37M
Student parameters
73.20%
Student accuracy
PROJECT 04
TinyML Gesture Recognition
A two-class gesture classifier built from custom six-axis IMU data, adapted CNN/LSTM pipelines, TensorFlow Lite conversion, quantization, and Arduino deployment.
6-axis
IMU input
2
Gesture classes
CNN/LSTM
Models evaluated
On-device
Arduino inference
MORE WORK
Additional Projects & Applied Research
These featured projects represent only a selection of my work. Additional projects span machine learning, natural language processing, computer vision, reinforcement learning, robotics, software engineering, and experimental research.
Technical Toolkit
Skills
Machine Learning & Deep Learning
Python
PyTorch
TensorFlow
Keras
Scikit-learn
Model Evaluation
Hyperparameter Tuning
NLP & Large Language Models
Hugging Face
BERT
T5
Llama
Mistral
Qwen
NLLB
M2M100
RAG
Local LLM Inference
Vision-Language Models
CLIP
BLIP
LLaVA
Multimodal Embeddings
Image-Text Representation
Computer Vision
OpenCV
OCR
YOLOv5
YOLOv8
R-CNN
Fast R-CNN
Faster R-CNN
Mask R-CNN
Model Optimization & Edge AI
Knowledge Distillation
Model Pruning
Quantization
TensorFlow Lite
TensorFlow Lite Micro
TinyML
Arduino Nano 33 BLE
Software Engineering & Tools
Java
C/C++
SQL
Git
Linux
Docker
Bash
PowerShell
LaTeX
Academic Background
Education
2023 – Present
Ph.D. Candidate in Computer Engineering, Artificial Intelligence & Robotics
Ferdowsi University of Mashhad · Mashhad, Iran
2024 – 2026
Doctor of Business Administration (DBA)
University of Tehran · Tehran, Iran
2022 – 2023
Master of Business Administration (MBA)
Ferdowsi University of Mashhad · Mashhad, Iran
2020 – 2022
M.Sc. in Computer Engineering, Artificial Intelligence & Robotics
Ferdowsi University of Mashhad · GPA: 3.9/4.0
Thesis: Hierarchical Solving of Euclidean Geometry Ruler-and-Compass Construction Problems Using Visual Knowledge Representation and Curriculum Learning.
2016 – 2020
B.Sc. in Computer Engineering, Artificial Intelligence
Ferdowsi University of Mashhad · GPA: 3.9/4.0
Contact
Let’s discuss AI/ML opportunities
Available for AI engineering, applied research, training, and technical collaboration.