Projects

Developed neural network methods and thermodynamic models to optimize the synthesis of 2D quantum materials, reducing computational predictions and experimental discovery time.

Designed and constructed a self-calibrating Raman Spectroscope with Qt-based GUI software, saving ~$500k.

Enhanced the MIT CosmicWatch with updated interactive features for science outreach.

Constructed automated DFT software for material categorization and value convergence, shortening computation times by up to 99%.

Designed a stock analyzing software that would utilize a sentiment analysis machine learning model to read stock news before the market opens to determine whether to trade stocks. The system was originally designed to integrate into the Robinhood trading software and handle daily trading activities by checking all stock values every 1 second. The software was designed to store and track each purchase, sell, and 1-second change of all stocks owned. Multiple model parameters were used for determining the limits for which to sell a stock at, resulting in some models averaging a ~200% annual return. Lower models averaged ~60% annual return.

Developed a system for dynamically controlling up to 3 CVD flow controllers. The system provided power to the flow constrollers, tracked the flow rates in real-time, and auto-adjusted flow rates to match desired settings. The system was also designed to allow for more complex changes and timings to flow rates, such as mathematical function derived flow rate curves. The system also tracked pressure alarms for the CVD system to allow for an automated emergency shutoff of all gas flow.

Created a bash-script that automates the installation of GPAW on ARM-based Apple Computers. The script installs GPAW in a Conda environment and installs all necessary dependencies. This was done before GPAW was officially supported on ARM-based Apple Computers. The script is now deprecated.

Developed a software for fitting Raman response curves for the SPC Raman Spectroscope. The software was used for high-throughput analysis of Raman response curves, reducing the time it took to analyze Raman response curves.

Created a Blender3D plugin that visualizes electron cloud densities for atoms and molecules. The plugin reads .CUBE files and converts them into OpenVDB files for visualization in Blender3D.

Wrote a resume builder that would take in a JSON file of the user's resume and the jobs that they're applying to and outputs a resume tailored to each job. The resume builder utilizes a large language model with a custom prompt and forced structures enforced by Pydantic.

Developed software to train and fine-tune large language models for the purpose of building a website chatbot that can be used to answer questions about the my resume and previous experience. More can be read about this on his <a href='https://pub.towardsai.net/how-to-train-a-language-model-on-your-resume-and-why-you-shouldnt-89bbddf9214d' target='_blank'>Medium page</a>