Experience

AISoccer Coach - Computer Vision Developer

- Present

  • Built a pipeline that groups tracked soccer players by team and identifies goalkeepers and referees. Compared 15 MobileNetV2 and ResNet50 models, with the best reaching 97.31% mean per-video-group clustering accuracy on a development benchmark of 5,581 crops.
  • Developed a video pipeline that recognizes 14 player actions and eight ball events using a fine-tuned Perception Encoder and temporal convolutions. The best ball-event model reached 76.2% macro-F1 on 180 labeled development clips.
  • Developed a pipeline that estimates a top-down 2D ball path from single-camera soccer footage by combining trajectory analysis, contact classification with a vision-language model, and player tracking.

PART AI - Computer Vision Developer

-

  • Built a prototype computer-vision pipeline for truck detection, tracking, and activity monitoring.
  • Trained YOLO models to identify phone use and smoking as workplace-safety events.
  • Prototyped a machine-vision method for monitoring the motion of three-piston industrial pumps.

ARIS Lab, University of Tehran - Research Assistant

-

Conducted applied research in robotics, computer vision, and embedded control under the supervision of Prof. Manouchehr Moradisabzevar.

Hand Puppeteer Robot

  • Designed and fabricated a 3D-printed robotic puppet integrating mechanical design, embedded control, and vision-based interaction.
  • Programmed a NodeMCU-based control system to process gyroscope measurements and translate gestures into robot movement.
  • Created and annotated a dataset of 2,000 images containing pose keypoints and bounding boxes.
  • Trained a YOLOv8 pose-estimation model and integrated its predictions with the robot for vision-based movement control.

Silkworm Robot

  • Contributed to the design and experimental evaluation of a modular, bio-inspired robot using passive magnetic adhesion for movement on ferromagnetic surfaces.
  • Designed and fabricated robot components using PLA, TPU, and Plexiglas, and developed an ESP32-based interface for remote operation.
  • Conducted locomotion experiments across different module speeds and frequencies to evaluate maneuverability and energy efficiency.
  • Contributed to the resulting research manuscript and experimental analysis.

Ganje -

-

Junior Front-End Developer

-

  • Developed a React and TypeScript interface that enabled repair technicians to access and manage smart parcel lockers.
  • Collaborated with front-end and back-end developers through Git-based workflows and code reviews to integrate locker-management features.

Mechanical Engineer

-

  • Designed sheet-metal smart lockers and produced CAD models and engineering drawings using SolidWorks and Onshape.
  • Used COMSOL simulations to evaluate locker components against environmental loads and vandalism risks.
  • Coordinated designs with manufacturing requirements to support fabrication and assembly.

Education

B.Sc. in Mechanical Engineering

University of Tehran, – · Tehran, Iran

GPA: Last two years: 3.48/4

Undergraduate thesis: Design of the Knee and Shoulder Joint Physiotherapy Device

Part of a collaborative whole-body rehabilitation device project; my contribution focused on mechanical design and analysis.

Thesis grade: 18.50/20

Supervisor: Dr. Alireza Daneshmehr

Publications

Journal article

A Wormlike Robot for Ferromagnetic Surface Inspection: A Novel System Inspired by Caterpillars

P. Parhami, M. H. Salehpour, H. Hamzeh, …, H. Moradi

IEEE Robotics & Automation Magazine (Early Access)

Conference paperOral presentation

Design of a Remote Controlled Puppet Robot Imitating a Manual Driven Puppet Using Deep Learning Pose Detection

H. Hamzeh, K. Shahroozi, A. Nabipour, …, H. Moradi

2024 12th RSI International Conference on Robotics and Mechatronics (ICRoM)

Selected for oral presentation at ICRoM 2024.

Under Review

Advancing Sustainable Pavement Design: A Data-Driven ML Framework for Predicting the Rheological Performance of Sasobit-Modified Binders

H. Hamzeh, P. Narimani, M. Dehghanpour Abyaneh

Skills

Vision

Object DetectionMulti-Object TrackingPose EstimationAction RecognitionUltralytics YOLOModel Interpretability

Honors & Awards

Fully Funded PhD Admission

USA · University of Connecticut

Admitted to conduct research on data-science applications in manufacturing.

Unable to enroll because the U.S. visa application was denied.

Winner, Third National Competition

Iran's National Elites Foundation

Recognized for developing a machine-learning model to identify and recommend suitable hydrogen adsorbent materials.

Top 2% Nationwide

Iranian University Entrance Exam (Konkur)

Ranked in the top 2% among more than 160,000 participants.

Certificates

Task-Oriented Bootcamp in MLOps

Quera

Completed with a perfect score

  • Built reliable data foundations with structured repositories, dependency management, SQL and PostgreSQL, idempotent ETL, schema validation, and pipeline tests.
  • Created reproducible training workflows with Prefect and managed experiments, artifacts, model packaging, evaluation, and promotion through MLflow.
  • Developed and tested batch and online inference services with FastAPI, Pydantic, structured logging, Docker, and Docker Compose.
View MLOps projects

Machine Learning Specialization

Coursera

  • Supervised Machine Learning: Regression and Classification
  • Advanced Learning Algorithms
  • Unsupervised Learning, Recommenders, Reinforcement Learning

References available upon request. Get in touch →