From September 2022 to April 2023, I worked on the Sensing team at Tesla as a Software Engineering Intern. My primary focus was on algorithm development for a new sensor, aimed at introducing new features and reducing costs. My daily tasks included training Machine Learning models on sensor inputs and occasionally supporting Signal Processing changes. The models I trained were primarily binary classification models used for detection and classification tasks. We iterated on these models to improve performance, achieving greater than 99.5% accuracy on features deployed to customers.

I conducted an initial assessment of a new detection feature, providing in-depth analysis to management to demonstrate the feasibility of the project. Based on the results I provided, the project received approval for full development and is scheduled for release to the customer fleet in the near future.

Additionally, I contributed to the development of the Python Machine Learning pipeline used to train various models, including Gradient-Boosted Decision Trees and Deep Learning models. I optimized our pipeline by addressing issues across vehicle firmware and machine learning components. This included automating data preprocessing steps, developing model reporting software to enhance reproducibility, and reducing model input feature size by 50% through quantization.

Furthermore, I led manual data collection efforts and supported data retrieval from our fleet, ensuring we had the necessary data to train our models. My work had implications across existing vehicle platforms on the road as well as vehicles yet to be released at the time.