I am a dedicated ML/Data Engineer with a proven track record of over 2 years, specializing in creating impactful products through the application of cutting-edge machine learning technologies. Throughout my journey, I have successfully contributed to diverse projects, showcasing my expertise in building robust data pipelines as well as delving into advanced areas such as Text-to-Speech and audio processing.
As an advocate of continuous learning, I am committed to staying current with the ever-evolving landscape of ML and data engineering. This mindset empowers me to embrace new challenges and implement novel techniques to drive project success. Beyond technical proficiency, I possess effective communication skills that facilitate seamless collaboration with cross-functional teams and stakeholders, ensuring that projects meet both technical and strategic objectives.
I'm passionate about NLP and audio processing tasks for the Belarusian language, like ASR, TTS, sentiment analysis, and language generation models.
SparkDrive: Igniting Data Insights in the Auto Sales Landscape
|
Engineered a project involving data extraction from av.by using an API, transforming raw data into a Data Lake, and processing it through Apache Spark.
Implemented seamless data orchestration by loading processed data into a Data Warehouse via Apache AirFlow, complemented by a Power BI report. Extended the project's reach by building a Flask-based web app for predicting car prices. |
Autovision: Unveiling Popular Car Classification through Images
|
Compiled and curated a dataset of 400,000 car photos, contributing to the model's exceptional accuracy. Fine-tuned a state-of-the-art car image classifier model using PyTorch, with F1-score of 0,9. Developed an intuitive FastAPI web app for effortless car classification from images. |
Urban Insights: Probing Minsk Apartment Market
|
Conducted detailed analysis and data processing of Minsk apartment data by parsing realt.by website information. Developed and autonomously executed the creation and exploration of predictive models for Minsk apartment prices, leveraging the acquired data. |
