Work Experience
“Choose a job you love, and you will
never have to work a day in your life."
~ Confucius
NVIDIA Self-Driving (Autonomous Vehicles)
Machine Learning Perception Intern
(May 2022 - August 2022)
Worked on improving classification performance for NVIDIA’s traffic sign models throughout the European Union, by converting simulation data to real-world data using zero reference images. Developed a custom zero-shot sim2real computer vision model using TensorFlow, OpenCV, and Python, by modifying and implementing various research papers. Improved classification accuracy on low-data image classes by average of ~10-15%, using my model to do sim2real image conversion which was used to boost performance on NVIDIA's next-generation core perception models.
Amazon Devices (Alexa)
Software Development Intern
(May 2021 - August 2021)
Developed backend software for a confidential and unreleased feature of Amazon Alexa. Worked primarily with Java, interfacing with various services such as AWS Lambda and Amazon DynamoDB.
FakeNetAI (startup)
Machine Learning Consultant (ML@B)
(January 2021 - May 2021)
FakeNetAI is a Berkeley SkyDeck startup that aims to detect synthetic media and deepfakes, in order to protect users from the dangers of these attacks through state-of-the-art fake video detection products. My role was the following: Developed and tuned various ML model architectures to distinguish between genuine vs. synthesized audio data found in deepfakes. Worked in a team of four students (through Machine Learning @ Berkeley), using PyTorch and Python to train models through Amazon EC2. Deployed the model onto a custom website to run inference on user audio input files.
DeWaste (startup)
Computer Vision Intern
(August 2020 - December 2020)
DeWaste utilizes cutting edge technology that collects real time data of food leftovers using computer vision. This data is then used to generate daily reports and actionable insights which can be used to reduce food waste by engineering the menu to adapt to customer preferences. I researched into deep learning models to classify food waste across various categories, performed preprocessing and analysis of data within public food waste datasets, and created documentation for future interns at DeWaste to continue diving into.
Lockheed Martin:
Space Systems
Machine Learning Intern
(June - August 2020)
Created a vision system using deep learning and computer vision to identify and label defects found within essential space-based hardware. Helped set up a new machine to automatically retrain/evaluate the model using GPU-accelerated ML with CUDA on an Nvidia DGX server. Trained the model up to 98% accuracy on collected data, and significantly increased model inference speed which reduced the company's defect detection time by 95%.
Lockheed Martin:
Enterprise Business Services
Software Engineering Intern
(June - August 2019)
Developed an application that allows authenticated Lockheed Martin employees to keep track of LM specific projects, replacing antiquated software no longer being supported by Oracle. Learned and applied various tools such as Angular, Node, Express, Bootstrap, HTML, CSS, and Oracle DB.