R&D Engineer - SoftServe
Project: Industrial IoT device development for visual data reading
Experience: My team and I developed a low-energy device with UI for calibration parts, which now can help factories automate receiving data from analogue gauges and make manual parts much more accessible. I was responsible for algorithm optimisation, creating the approach of reading the rotated gauges, benchmark dataset creation, recording and postprocessing, and algorithm benchmarking.

Project: Efficient Generative AI
Experience: During this project, my main research was the large language models optimisation and its impact on model accuracy and inference. Furthermore, I worked with AMD GPU for the pre-sale from Dell to find out how the different structures of GPUs impact the model inference after optimisation.

Project: Automatic Speech Recognition
Experience: The project's main idea is to develop a user-friendly pipeline to analyse childish speech and detect pronunciation issues. I was responsible for working with audio-to-text models and their usage in the pipeline and model training for the interaction with phonemes. Also, we were cooperating with the team from Nvidia to develop a similar pipeline using Nvidia Audio Effects.

Data Scientist/Analyst - The Parker Avery Group
Project: The Hypercube
Experience: I am responsible for developing the interaction between the user request to the AI Agent and Forcasting System, and further feedback creation. It includes response formulation and providing a necessary statistics in the form, which would be easy to understand for people, who works in retail sphere.

Project: Item matching
Experience: I developed the pipeline of computing the similarity of the items/products depending on their description, price and extracted size out of the description. For the embeddings I used the OpenAI API, which provided vectorisation of the necessary data with good accuracy.