LITSLINK is software development company from the US that unleashes the power of artificial intelligence with AI business solutions. Our goal is to make this advanced technology accessible and help companies streamline their operations with smart automation tools. 𝗥𝗘𝗦𝗣𝗢𝗡𝗦𝗜𝗕𝗜𝗟𝗜𝗧𝗜𝗘𝗦> Communicating with the clients and producing project estimations> Developing and optimizing machine learning models and pipelines> Gathering and processing data required for training> Mentoring trainees> Providing consultation and leading other developers on NLP tasks> Deploying the models as APIs or simple UIs𝗔𝗖𝗛𝗜𝗘𝗩𝗘𝗠𝗘𝗡𝗧𝗦• Gathered 300+ hrs of audio and fine tuned a custom model based on the encoder of OpenAI's Whisper for grading English learners' speech pronunciation and fluency.• Prepared the dataset and fine tuned a gpt-4o-mini model to generate personalized emails, which allowed the company to switch from using costly gpt-4o.• Set up an automation for finding prospects for cold email outreach, which got the team caught up on about half a years work• Improved the RAG implementation of a tax research chatbot, by adding metadata filtering, reranking and agentic request processing.

Machine Learning Engineer at Alcoras. They provide technical support for business automation software for multiple companies and universities such as National Aviation University, Taras Shevchenko National University of Kyiv, etc., but are also expanding into other related niches. My role was to assist their R&D team in developing efficient machine learning models.𝗥𝗘𝗦𝗣𝗢𝗡𝗦𝗜𝗕𝗜𝗟𝗜𝗧𝗜𝗘𝗦> Developing and optimizing machine learning models> Gathering and processing data required for training> Synthesizing additional data> Monitoring equipment during the training process> Creating extensive reports on the progress𝗔𝗖𝗛𝗜𝗘𝗩𝗘𝗠𝗘𝗡𝗧𝗦• Set up the server to use GPU, which increased the training speed by 4 times on average.• Manually gathered 400+ images of receipts on phone cameras and later generated 10,000+ synthetic images by varying rotation, brightness and blur.• Created a supervised learning model for reading the "total" value on the receipt, which had 96% accuracy and could process images at a rate of 10+ fps.

Full Stack Developer at Data Computing Center Of Taras Shevchenko National University of Kyiv. They maintain and improve all of university's computing and network infrastructure and resources. My role was to assist the software development team headed by the Chief engineer.𝗥𝗘𝗦𝗣𝗢𝗡𝗦𝗜𝗕𝗜𝗟𝗜𝗧𝗜𝗘𝗦> Developing software solutions and applications> Maintaining server infrastructure> Responding to customers' problems at the helpdesk> Refining applications in accordance to customer feedback> Training new hires𝗔𝗖𝗛𝗜𝗘𝗩𝗘𝗠𝗘𝗡𝗧𝗦• Developed an admin dashboard for infopacket.knu.ua, which allows representatives from all faculties to manage their parts of the website.• Remade old website of the Computer Engineering department ce.knu.ua using modern tools and deployed it using an automated CI/CD pipeline.• Developed a website template for future use by the company• Using the template, created a login website for future Single Sign-On system.