NLP tasks: dataset preparation, cleaning, tokenizer development, causal (GPT-2) and masked (RoBERTa) language model training and fine-tuning, pipeliningStatistical modeling: generalized linear multi-level mixed effects models with STANInterface development: study design, transcriber web-app with jsPsych, JATOS

I greatly appreciated having a second year to continue my work at ICON Whitesboro. Ultimately, I continued my work from my previous internship, working on integrating liquid handlers. I developed and GLP-validated Hamilton Liquid Handler protocols for everyday use in lab tasks. For future programmers, I wrote libraries for usage of integrated sample decappers, and sample barcode readers in the automation workflow.In addition to my work with automation, I also worked with Project Managers on the creation of a tool for streamlining report generation, and I worked with scientists to create a tool to validate plate layouts. Finally, I began work on a JavaScript-based curve weighting regression solver tool for analyzing heteroskedasticity.

I worked with ICON on integrating a series of Hamilton automated liquid handlers. These will be used to automate repetitive and time-consuming tasks in the lab, such as preparing standard curves and QCs. It is important to the bioanalytical lab to have a high degree of precision and accuracy, and so I ran a series of tests to determine the limitations and best practices for the liquid handlers, to maximize throughput along with trueness. Along the way, I learned the basics of ELISA and other immunoassays, as well as the principles and extraction techniques of LC/MS-MS. I also created several Excel tools to better facilitate interactions between the scientists and the robots. At the end of my internship, I gave a presentation on my findings to management and senior lab staff.