The first Machine Learning (ML) developer in the company, I coordinated the development and launch of an AI assistant based on Large Language Models (LLMs) for the students. I advised cross-functional teams on the processes of creating various ML projects and participated in their implementation, focusing on ML tasks. I set up the inference of LLM models and developed quality assessment and monitoring pipelines (dashboards) from scratch. I was engaged in prompt engineering and quality improvements according to developed scenarios. With the help of assessors, I labeled data for quality assessments and model fine-tunings. Interviewed new ML developers to expand the team.

I develop and teach project-based data science courses for high school students. Cover topics on math for data science, data analysis fundamentals, classic ML and deep learning algorithms. Designed and taught 8 course iterations for 100+ students total, during summer and winter bootcamps as well as regular group and private classes.

Developed a detailed educational program for the interactive text-based "Math for Data Analysis" course. Developed a curriculum including topics, lessons, and educational outcomes. Advised the authors on the content plan and some advanced data science topics. Participated in content writing, have written some of the course lessons. Currently, I continue to be involved in supporting the course.

Conducted research and experiments at the intersection of machine learning and neuroscience. Some topics are:- Applications of generative neural networks to model 3D fMRI brain data - Analysis of visual cortex neuronal activity in monkeys with machine learning methods to study short-term memory- Applications of machine learning methods for pen trajectory prediction using electromyography recorded from the wrist - Evaluation of human biometrics (heart rate, SpO2) from optomyography using Bayesian filters

Research in 3D computer vision in the self-driving cars department. Developed efficient neural network approaches in point cloud segmentation.

Developed a working prototype of a web application for real-time OCR in a browser using OpenCV and WebAssembly. Optimized document detection algorithm, increased FPS of the application on mobile devices by 10+.

Conducted research and testing of audio speech analysis solutions: ASR, VAD, speaker diarization. Developed speech rate estimation pipeline, integrated it into Microsoft Azure service.