Redzhep Mehmedov
Verified Expert in Engineering
Computer Vision Expert and Developer
Munich, Bavaria, Germany
Toptal member since January 20, 2026
Redzhep is an expert in computer vision and reinforcement learning. He has led teams in robotics companies for four years. He has crafted numerous AI applications for diverse global scenarios, showcasing strategic innovation. Redzhep is a valuable addition to any team, bringing extensive experience and expertise.
Portfolio
Experience
- Python - 9 years
- Deep Learning - 8 years
- Computer Vision - 8 years
- Object Detection - 7 years
- PyTorch - 7 years
- AI Project Management - 6 years
- Reinforcement Learning - 5 years
- Leadership - 4 years
Preferred Environment
PyCharm, Slack, PyTorch, Linux
The most amazing...
...thing I've developed is a robust policy training framework with my team for long-horizon robotic control tasks utilizing solely image and proprioceptive data.
Work Experience
AI Team Lead
Sunrise Robotics
- Developed and implemented the initial computer vision pipelines for MLOps.
- Created and implemented RL policies for long-horizon robot control tasks such as pick and place, extract, lift, and more, using NVIDIA Isaac Lab.
- Built and implemented the end-to-end configurable synthetic data generation pipeline using NVIDIA Isaac Sim.
- Developed and prototyped a zero-shot recognition platform's AI modules.
- Researched and improved a SOTA monocular camera-based 6D pose estimation method to make it work with high accuracy for industrial robotic operations.
- Developed an end-to-end hiring strategy and hired 10+ engineers to different teams in the company.
- Managed and guided the team to rapidly deploy 6+ robots to different customers with different use cases.
Lead AI Engineer & Solution Architect
Kolektor companies
- Architected a full computer vision and deep learning pipeline to handle active data gathering, auto-annotation, data analytics, algorithm construction, deep model training, optimization, automatic model conversion, and model deployment.
- Designed and managed a semi-automated end device application development, testing, integration, and iteration pipeline.
- Created high-level solutions for any given new problems or project proposals and helped the team build the end applications that can be used within the robot.
- Architected the Sim2Real pipeline and managed the experimentation process. Built the initial smart translation module for the robots.
Lead AI Engineer
OTSALA AI
- Architected and managed the development of a full computer vision and deep learning pipeline to handle active data gathering, auto-annotation, data analytics, algorithm construction, and deep model training and optimization.
- Implemented an approach-agnostic solution for a realistic dynamic job shop scheduling problem. Achieved top results using the improved versions of the PPO and SAC algorithms on top of adaptive action masking and observation normalization strategies.
- Developed a time-series data processing, analysis, prediction, and evaluation system. Implemented an ensemble forecasting algorithm using deep learning, Bayesian optimization, and gradient boosting algorithms.
- Used mixed-transformers (Bert, ViT, and MAE) to build custom CLIP, XCLIP, or SSL-based recommendation engines. Some of the approaches were also used to perform zero-shot classification and object detection.
Computer Vision & Deep Learning Engineer
SKIDATA GMBH
- Adjusted a deep detection model’s architecture by following the state-of-the-art neural network to make the model more accurate in recognizing overlapped objects and scalable with data.
- Accomplished a complete examination of the impact of the pre-processing algorithms on deep classification algorithms and developed a configurable pre-augmentation module.
- Developed a fully unsupervised image deduplication algorithm by combining deep learning and classical machine learning algorithms.
- Prototyped a semi-automatic synthetic dataset creation algorithm that generates a high-quality multi-purpose annotated dataset to minimize data requirements from a customer and reduce the operational costs, such as long capturing sessions.
Experience
Recurrent PPO Plays Doom
https://github.com/redzhepdx/RecurrentPPO-CppMultiMediaSearch
https://github.com/redzhepdx/MultiMediaSearchPLSegmentationPipeline
https://github.com/redzhepdx/PLSegmentationPipelineEducation
Bachelor's Degree in Computer Engineering
Yildiz Technical University - Istanbul, Türkiye
Certifications
Fundamental of Accelerated Computing with CUDA Python
NVIDIA
Deep Reinforcement Learning Nanodegree
Udacity
Convolutional Neural Networks
Coursera
Introduction to Game Development
Coursera
Skills
Libraries/APIs
PyTorch, NumPy, Scikit-learn, OpenCV, Pandas, Hugging Face Transformers, PyTorch Lightning, TensorFlow
Tools
Visual Language Models (VLMs), You Only Look Once (YOLO), PyCharm, Slack
Languages
Python, C, C++
Paradigms
Synthetic Data Generation, ETL
Platforms
Amazon Web Services (AWS), Docker, Ubuntu, Linux, NVIDIA CUDA
Storage
Amazon S3 (AWS S3)
Frameworks
Unity
Other
Deep Learning, Computer Vision, Algorithms, Data Structures, Reinforcement Learning, Machine Learning Operations (MLOps), AI Project Management, Semantic Segmentation, Convolutional Neural Networks (CNNs), Object Detection, Artificial Intelligence (AI), Machine Learning, Image Processing, Data Science, Pose Estimation, Vector Databases, Recommendation Systems, IT Management, Custom Models, Multimodal Models, Hugging Face, Computer Vision Algorithms, Actor-critic Methods (A2C, A3C), Transformer Models, Decision Trees, Dimensionality Reduction, Time Series Analysis, K-means Clustering, Linear Regression, Logistic Regression, Random Forests, Image Analysis, Image Segmentation, 3D Pose Estimation, Video Analysis, Predictive Modeling, Data Analysis, Leadership, Time Series Forecasting, GPU Computing, Stable Diffusion, MLflow, Time Series, Video Processing, Video & Audio Processing, Retrieval-augmented Generation (RAG), Prompt Engineering, API Integration, Combinatorial Optimization, RAG Pipelines, Large Language Models (LLMs), AI Agents, RAG Systems, LangChain, Generative Pre-trained Transformers (GPT), Gradient Boosting, Motion Capture, Vehicle Tracking Systems, HRL, Energy, Reinforcement Learning from Human Feedback (RLHF), Light LLMs, NVIDIA TensorRT, Deep Reinforcement Learning, Game Development, PyCUDA, Numba, Instance Segmentation, Image Classification, Metric Learning, Support Vector Machines (SVM), Financial Markets
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring