Viktor Turchenko Email & Phone Number

Ontario, Canada

Ontario, Canada

• Benchmarked tensorization methods for machine learning models, applying them to time-series forecasting for Andretti Motorsports.
• Modified and enhanced custom models by adjusting weight flow and code structure, leading to improved performance.
• Compared performance of custom JAX and PyTorch models for time-series prediction, enhancing custom model efficiency.
• Developed a robust metric system to evaluate model performance; improved predictive accuracy for the Andretti Motorsports use case.

Ontario, Canada

• Fine-tuned LLMs to specifically target and improve processes in logistics and document processing, resulting in enhanced data processing and operational efficiency.
• Leveraged the deployment of models in practical work environments, ensuring they are robust and adaptable to meet the evolving demands of the business.Developed and deployed backend APIs integrating machine learning.
• Performed code maintenance and created test cases for model evaluation, implementing performance metrics tracking on Microsoft Azure for monitoring and optimization.

London, Ontario, Canada

• Collected a mental health state dataset for a classification problem
• Developed state-of-the-art natural language processing models (RNNs, CNNs and Transformers) to accuratelyclassify human mental states, leveraging advanced techniques to enable high-performing predictive models
• Created a sophisticated mental health state classifier by integrating diverse input types including text and numerical features, resulting in an accurate predictive model using one-fold and K-fold cross-validations
• Delivered a production version of the mental health state classifier for in-app usage

Toronto, Ontario, Canada

• Designed DNNs and CNNs (regression and classification) from scratch and with transfer learning to predict blood pressure with Tensorflow
• Used signal processing libraries (internal and external) to process PPG, ABP, and ECG data from multiple datasets
• Created a python package for extracting signals from datasets with multiprocessing features
• Replicated a blood pressure predicting technique from a scientific paper, as well as the data generation method
• Created reports on Confluence, uploaded code to BitBucket, created tickets on Jira