● Open to New Opportunities

Mohammadreza (Hamid) Matiny

Machine Learning Engineer & MLOps Specialist

Experienced Python developer specializing in building scalable machine learning systems, MLOps pipelines, and full-stack applications using PyTorch, FastAPI, Docker, and cloud technologies.

View My Projects Let's Connect
Hamid Matiny

About Me

Results-driven Machine Learning Engineer and Python developer with over 6 years of experience building end-to-end machine learning solutions, MLOps pipelines, and scalable web applications.

Currently at Torc Robotics, I work with large-scale LiDAR and camera data to improve autonomous truck perception systems. In parallel, I develop production-grade ML pipelines using PyTorch, FastAPI, Docker, and modern deployment strategies.

My expertise spans model development, performance optimization, API development, and deploying reliable systems from experimentation to production.

Featured Projects

Cuda Optimization (MPS)

Achieved 2–4× speedups and ~60% memory reduction on Apple M3 hardware using mixed precision, torch.compile, and gradient checkpointing.

View Repository →

Itinera

Built an intelligent route-planning engine that processes real-time API data and multi-constraint user profiles to deliver optimized, interactive travel itineraries.

View Repository →

Sentinel Ray

Distributed camera telemetry ingestion with Ray Core, Pandera QA validation, statistical drift detection, and automated incident orchestration.

View Repository →

Hydra Data Factory

End-to-end data engineering platform for autonomous-vehicle (AV) telemetry: mock fleet ingestion, fault-tolerant ETL, Pandera data contract validation, containerized orchestration, Terraform-managed AWS infrastructure, and Glue-synced Parquet lakehouse storage.

View Repository →

Vanguard Telemetry Monitor

A production-pattern vehicle telemetry simulation and observability platform built in four incremental phases. Vanguard generates realistic fleet telemetry, injects production-like anomalies, exposes Prometheus metrics, visualizes dashboards in Grafana, automates deployment and recovery with Bash, and validates every change through pytest and GitHub Actions CI. Develop on macOS, Windows, or Linux — the core workload always runs inside a containerized Linux environment via Docker, mirroring how real fleet infrastructure is operated in the field.

View Repository →

Object Detection

Developed a real-time object detection system using YOLOv8 and PyTorch, achieving high accuracy and efficiency in various environments.

View Repository →

Vision Transformer (ViT) Implementation

A high-performance FastAPI service that classifies images into the 10 CIFAR-10 categories. This project evolved from a custom-trained ResNet-18 to a Vision Transformer (ViT) using Transfer Learning from Hugging Face.

View Repository →

Task Master

A production-ready FastAPI REST API with PostgreSQL, Docker, and GitHub Actions.

View Repository →

Multi-Modal Object Detection

A comprehensive object detection system that fuses data from multiple sensors (Radar, LiDAR, and Camera) to detect and annotate objects for autonomous driving models.

View Repository →

ADAS Next Best Action Recommendation

This repository contains an ADAS-style predictive model with sequential and ensemble logic for next best action recommendation.

View Repository →

LSTM & Attention & Transformers

The evolution of Attention: RNNs, LSTMs, and Transformers. Implementations, fine-tuning, and visual attention mapping.

View Repository →

ML & DL Fundamentals

A collection of fundamental concepts and implementations in machine learning and deep learning.

View Repository →

MLOps & Deployment

A production-ready FastAPI application that serves a Hugging Face Shakespeare language model

View Repository →

MLOps & Deployment of CIFAR-10

A high-performance FastAPI service that classifies images into the 10 CIFAR-10 categories(v2.0).

View Repository →

FastAPI BERT Inference Docker Container

This project containerizes a FastAPI application with BERT (DistilBERT) model inference using Docker. The application provides sentiment analysis capabilities through REST API endpoints.

View Repository →

Anti-Money Laundering and Counter-Terrorist Financing (AML/CTF) System

This project implements a comprehensive machine learning system for detecting suspicious financial transactions related to money laundering and terrorist financing. The system uses advanced data analysis techniques and multiple ML models to identify high-risk transactions and patterns.

View Repository →

Technical Skills

Machine Learning & Deep Learning

  • PyTorch, YOLOv8, Transformers, Vision Transformers
  • Object Detection, Sensor Fusion, Semantic Segmentation
  • Model Optimization (AMP, torch.compile, MPS)

MLOps & Deployment

  • FastAPI, Docker, Docker Compose
  • MLflow, CI/CD, GitHub Actions
  • Model Serving & REST APIs
  • Cloud Deployment (GCP, AWS)

Software Engineering & Backend

  • Python (Expert), SQL, Pandas, FastAPI, Flask
  • PostgreSQL, Data Pipelines, ETL
  • Full-Stack Development (React + FastAPI)
  • System Design & Performance Optimization

Experience

Data & Quality Assurance Engineer – ML Pipelines

Torc Robotics • 2024 – Present
  • Worked with perception and simulation teams to ensure high-quality LiDAR & camera data for production ML models.
  • Supported data pipelines and performed quality assurance that improved downstream model reliability.

Data Scientist

ReIAI • 2019 – 2024
  • Developed and deployed real-time object detection models and predictive systems achieving high accuracy.
  • Built automated data pipelines using Docker, FastAPI, and MLflow.
  • Designed and maintained scalable backend systems and APIs.

Software Developer

Saeidi Freight Institution • 2014 – 2019
  • Built scalable web applications and REST APIs using Flask and Python.
  • Optimized application performance and implemented data-driven solutions.

Get In Touch

Open to new opportunities in MLOps, Machine Learning Engineering, and Python Backend Development.

hamidmatiny@gmail.com
Lets get to work