Aydar Mynbay
Verified Expert in Engineering
Machine Learning Developer
Almaty, Almaty Region, Kazakhstan
Toptal member since November 27, 2024
Aydar is a machine learning engineer with deep expertise in reinforcement learning, demand forecasting, and recommendation systems. He has worked with global companies such as Adidas and PUBG and co-founded an AI R&D startup focused on optimizing last-mile delivery. Aydar is also skilled in back-end development, MLOps, and data engineering. He holds a degree from the Korea Institute of Science and Technology (KAIST), ranked among the top 50 universities globally by QS World Rankings.
Portfolio
Experience
- Machine Learning - 5 years
- Python - 5 years
- PyTorch - 4 years
- Reinforcement Learning - 4 years
- Spark - 3 years
- Deep Reinforcement Learning - 3 years
- TensorFlow - 2 years
- Recommendation Systems - 1 year
Availability
Preferred Environment
Ubuntu, PyCharm
The most amazing...
...thing I've achieved is co-founding an AI startup with $1+ million of venture capital investments.
Work Experience
Senior Machine Learning Engineer
Adidas
- Developed a temporal fusion transformer model for demand prediction, which decreased mean absolute error by 10% compared to a linear model.
- Implemented Bayesian hyperparameters optimization and distributed learning with Amazon SageMaker and PyTorch Lightning.
- Designed and developed an ETL pipeline for pricing models using PySpark and Airflow. Implemented CI, unit, integration, and data quality testing.
- Implemented simulation based markdown optimization algorithms.
Machine Learning Engineer | Co-founder
Solai Inc.
- Outperformed traditional approaches by 5-20% by researching and developing deep reinforcement learning algorithms for solving the vehicle routing problem (VRP), including creating a discrete event simulator for pickup and delivery issues.
- Designed and developed a transportation management system back-end API with Go Fiber and PostGIS and implemented a CI/CD pipeline with GitHub Actions.
- Reduced expenses by 60% by developing and training models using XGBoost and graph neural networks to lessen distance matrix API requests. Used Amazon Sagemaker to implement the ETL and training pipeline.
- Improved on-time delivery from 50 to 90% and reduced average delivery time from 70 to 50 minutes by designing and creating a simulated annealing-based real-time dispatcher for on-demand grocery delivery. Built XGBoost-based models for ETA prediction.
- Researched and developed meta-heuristic algorithms to solve various black-box optimization problems. Configured and implemented VRP and linear programming solvers such as jsprit, VROOM, OR-Tools, LKH-3, Gurobi, and CP-SAT.
Machine Learning Engineer
Aitu
- Designed and developed the ETL pipeline for a video-sharing platform's recommendation system.
- Increased the click-through rate from 4 to 6% by developing collaborative filtering and content-based recommendation systems in an A/B testing environment.
- Developed a deep convolutional model for detecting not-safe-for-work content.
Machine Learning Engineer
PUBG (trademarked by KRAFTON)
- Created an image classification model and a weakly supervised object localization model to detect extra sensory perception cheat occurrences on users' gameplay screenshots—achieving 95% accuracy and above 0.5 Intersection over Union value.
- Developed a time series classification model for auto-aim and recoil-control cheating detection—achieving 80% recall and 95% precision on the unbalanced dataset.
- Handled the logic design, data analysis, and development of the behavior tree for a data-driven AI bot for PUBG PC and console—outperforming the previous bot and replacing it in later updates.
Experience
Extended Mean-field Inference Theory and RL Applications to NP-Hard Multi-robot/machine Scheduling
https://grlplus.github.io/papers/91.pdfEducation
Bachelor's Degree in Industrial and Systems Engineering
Korea Advanced Institute of Science and Technology (KAIST) - Daejeon, South Korea
Certifications
Natural Language Processing Specialization
DeepLearning.AI | via Coursera
Skills
Libraries/APIs
PyTorch, TensorFlow, PyTorch Lightning, PyTorch Geometric (PyG)
Tools
Amazon SageMaker, Gurobi, PyCharm, Apache Airflow, Git
Languages
Python, SQL, C++, Go, C, Java
Frameworks
Spark, Unreal Engine 4, SimPy, gRPC, Django
Platforms
Apache Kafka, Docker, Ubuntu, Amazon Web Services (AWS)
Storage
PostgreSQL
Other
Machine Learning, Artificial Intelligence (AI), Data Science, Reinforcement Learning, Operations Research, Mixed-integer Linear Programming, Stochastic Modeling, Recommendation Systems, Deep Reinforcement Learning, FastAPI, Research, Graph Neural Networks, GitHub Actions, Natural Language Processing (NLP), Transformers, Logistics, Optimization Algorithms, Demand Forecasting
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