Steven Palayew Email & Phone Number

Toronto, Ontario, Canada

Research on LLMs + Multimodal Learning in the biomedical domain.

Toronto, Ontario, Canada

• Research in grounding LLMs, applications of LLMs in the biomedical domain.• Supervised by Dr. Bo Wang.

• Brought a project recognized by the United Nations for innovatively leveraging AI in sustainable agriculture from an initial proof of concept, to a minimum viable product. This was accomplished firstly through moving to a constrained differential evolution based approach, which improved yield optimization performance by approximately 40-50%, as well as ensured results were realistic, and deterministic. In addition, deployed the project as a web app in order to gather customer feedback. Technologies used for this work include Python, SciPy, XGBoost, Docker, Streamlit, and CSS. • Core contributor to a research project where a novel backbone was integrated into RetinaNet.• Research supervised by Dr. Alexander Wong.

• Using transformer models, developed NLP tools for identifying PubMed articles with biomolecular interactions of potential interest. • Preprocessed training and test data for these models, using confident learning and the Cleanlab library to help quantify and rectify hundreds of labelling errors. • Supervised by Dr. Gary Bader.

Toronto, Ontario, Canada

• Using R and Python, identified important biomarkers to focus on when researching improved methods of testing for and treating granulomatosis with polyangiitis (GPA). Based on a thorough literature review process, calculated feature importance using ElasticNet and random forest models, aggregated results of this using Borda’s method, and ensured these models performed well using nested cross-validation.• Supervised by Dr. Katherine Siminovitch.

Montreal, Quebec, Canada

• Using Python with the Pandas, Gensim, NLTK, XGBoost, and Scikit-Learn libraries, developed a proof of concept for a completely automated record linkage system which employed cost-sensitive machine learning algorithms. This system could be used to significantly improve the speed at which data on healthcare professionals could be provided to clients by eliminating the need for manual intervention. • Developed a pipeline to query an Amazon DynamoDB database and pre-process training and test data for this system. • Improved system scalability through distributed computing using PySpark. • Presented results of my research to developers and senior leadership across Canada through information sessions that I organized and hosted.

Toronto, Ontario, Canada

• Using Python with the Keras, TensorFlow, Pandas, and Scikit-Learn libraries, developed neural network models to improve the accuracy of automated genotype calling. Reduced error rate from previously used software by over 30% (tested using k-fold cross validation). • Developed a VBA program to automate quality control of genomics data. • Generated visual displays of genomics data in Microsoft Excel that were presented to various stakeholders.• Optimized procedures for automated fluid handling to improve the cost effectiveness and significantly increase the speed of related procedures.

Ottawa, Ontario, Canada

• Assisted in the development of a new database to track Canadian researchers’ applications for blood products. Drafted guidelines for Canadian researchers to use the new, more user-friendly system. • Assisted in web design to improve the aesthetic appeal and overall usability of web services associated with Canadian Blood Services.

Msc, Computer Science

Supervised by Dr. Gary Bader, Dr. Bo Wang

Bachelor Of Applied Science - Basc, Mechatronics Engineering | Computing Option

Activities and Societies: Tron Mentor Chats: Peer Mentor