Data Engineer
2020 - 2021Let's Enhance- Used a Triton Inference Server to significantly speed up neural processing up to two times.
- Maintained the production on Google Kubernetes Engine (GKE) and handled support requests and any detected errors.
- Supported and improved CI/CD pipelines and made changes to suit them for ever-changing needs.
- Developed custom client solutions, including image-CDN enhancement.
Technologies: Triton Compute, Python, TensorFlow, PyTorch, PostgreSQL, ClickHouse, Google Cloud, Google Kubernetes Engine (GKE), Kubernetes, Machine Learning, Git, Data Engineering, SQL, Data Architecture, Cloud Architecture, Business Intelligence (BI), Metabase, AWS, Relational Databases, Helm, Terraform, GitLab CI/CD, CI/CD Pipelines, Business Intelligence (BI) Platforms, Data-driven Dashboards, DockerFounder
2019 - 2021Smart Home Solutions- Completed customers' orders to develop smart, custom-home appliances, custom PCBs, and software.
- Completed a remotely controlled motor project and developed a Flutter app to control it.
- Found and negotiated with customers interested in smart home solutions.
Technologies: Flutter, iOS, Android, KiCad, Autodesk Fusion 360, CNC Routers, PCB DesignData Engineer
2018 - 2020Quartesian- Created a model to remove the background from images. Wrote a RabbitMQ worker to process the images to make the model available for mobile apps.
- Developed a financial-data webserver able to accept financial statements and perform queries upon them. Used Neo4j to store interconnections.
- Created a facial-recognition system able to remember up to 150,000 unique people and classify them with a low-error rate. I used different types of workers, a few databases, and a few data sources.
- Wrote a module for Android and iOS apps to have onboard-neural networks able to detect, crop, and encode faces to significantly speed up the back-end processing.
- Developed load testing and profiling tools to identify issues and ensure consistency across different devices and services.
- Deployed services using DigitalOcean and wrote custom deployment scripts.
- Developed a custom dashboard to monitor ETL tasks and face recognition data processing.
- Created a service to visualize the WHO Drug database to quickly find medical data by drugs and symptoms.
Technologies: Python, Deployment, Android, iOS, RabbitMQ, Redis, PostgreSQL, Machine Learning, Aiohttp, Data Engineering, SQL, Data Architecture, Data Visualization, Grafana, Data Warehousing, CouchDB, Relational Databases, Dimensional Modeling, MongoDB, NoSQL, Neo4j, React, JavaScript, Webhook, Web Dashboards, Dashboards, React Native, Data Analytics, ETL, Data Validation, Data-driven Dashboards, Docker, Pandas, Data PipelinesMachine Learning Engineer
2018 - 2018Simporter- Used Twitter API to crawl and collect all of the entity's mentions.
- Classified texts and scored how customers are happy with the products they buy, using sentiment analysis models.
- Used the Baas diffusion model to develop a model to score the products, according to their mentions in the social media, and predict sales.
Technologies: Python, Scikit-learn, Twitter API, Sentiment Analysis, Natural Language Processing (NLP), Data Visualization, NoSQL, MongoDB, Redis, Dimensional Modeling, Data Analysis, Azure, Data Validation, PandasMachine Learning Engineer
2016 - 2018ARVI VR- Developed the system to recognize web pages according to IAB's content taxonomy, using NLP techniques, including embedded learning, clustering, word2vec, and doc2vec.
- Used and managed the in-house GPU cloud based on OpenStack.
- Wrote an algorithm to compare facial expressions and developed the app with a distributed back end to read facial expressions on a scale.
- Helped to organize the machine-learning weekend course.
- Used the Julia language to write a high-speed implementation of the OPTICS data-mining algorithm to process a data lake.
Technologies: Python, TensorFlow, SpaCy, OpenStack, Data Analysis, Aiohttp, Julia, Data Lakes, MySQL, PandasResearch Assistant
2016 - 2016Institute of Mathematical Machines and Systems Problems NAS of Ukraine- Developed an aerial-imagery classification neural network for unmanned-aerial vehicles.
- Searched patents to find the related algorithms and reproduced them to compare with our methods.
- Created the images dataset to be used for neural-network training and validation.
Technologies: Python, MATLAB, Machine Learning, Patents, Linux, Ubuntu, Data Analysis