Air Quality- Developed deep learning model with PyTorch to predict global temperature with test accuracy of 95%- Queried and data cleaned an internal database using MySQL to extract features closest to temperature- Leveraged features such as air temperature, wind speed, and solar radiation to build the model- Built a model with 8% accuracy greater than the best existing modelSurface Temperature Acceleration- Extracted PM2.5 (particulate matter) data via the OpenAQ API then added it to a existing MySQL database- Preprocessed hourly PM2.5 data to calculate daily, monthly, and seasonal trends- Trained Random Forest (best model) to predict PM2.5 values using aerosol optical depth, CO2 and NO₂- Identified where air pollutants are greatest to prepare for situations with unbreathable air

- Developed C++ library to synchronize cameras to capture better quality data for COVID-19 detection- Created CNN based model to reduce bias in detecting the heart rate of people with darker skin tones- Detected heart rate using an iPhone camera and flashlight using fast fourier transform with 96.7% accuracy- Created iOS app for simultaneous capture with wide and ultra-wide cameras for predicting blood pressure- Developed eye tracking algorithm for HTC Vive and Oculus Quest 2 to aid Mixed-Reality experiences

I taught the ACM AI Advanced Workshop, which introduced students to deep learning and neural networks: CNN, LSTM, and RNN with a guided MNIST classifier project at the end.

- Analyzed precipitation data to observe patterns in rainfall that appeared in the tropics- Trained K-means clustering model with 92% accuracy to observe precipitation patterns- Detected precipitation patterns which can indicate the onset of a hurricane or cyclone earlier- Data mined precipitation data that I queried from a MySQL database to calculate seasonal trends