I'm exploring deep learning architectures for audio synthesis using Large Audio Language Models (LALMs), with the exciting goal of advancing AI capabilities in neural audio generation and patch synth sound design. Currently, I'm working with TensorFlow and PyAudio to study DDSP-based models and understand their application in generating (and attacking) audio. I'm beginning to explore the role of GANs in refining audio synthesis techniques, and am enjoying learning about sound realism through this exciting lab.

I focused on enhancing a machine learning model for the AI&D product team to improve network performance and security monitoring through anomaly detection for business customers. This experience also deepened my expertise in data engineering, refining my ability to optimize AI-driven predictive analytics for operational efficiency. Additionally, I collaborated with my cohort to architect cloud-based solutions for a Fortune 50 client, leveraging AI and cloud security frameworks to mitigate IoT ransomware vulnerabilities using the MNS product portfolio.

I'm working with an Entrepreneurs@Berkeley team contracted on a 14-week project focused on researching the product lifecycle and go-to-market strategy for one of LinkedIn's key products in international markets. Our work involves evaluating transformation tools to enhance potential platform relevance for next-generation users. We're developing and pitching strategic feature reworks and new implementations aimed at capturing emerging audiences and driving user engagement, cool stuff!

I was a Project Manager for the 2024 project cycle with Entrepreneurs@Berkeley, a Haas affiliated consulting and entrepreneurial organization. Through the E@B ecosystem, I've begun working on a social media startup and completed a technical consulting project with Theya.

I'm part of a Berkeley-based team developing Pour Decisions, a social-safety application designed to promote a safer and smarter partying culture among college students. The app leverages cloud-based facial recognition and deep learning to integrate pymremetrics for intoxication monitoring. Our goal is to create a proactive safety tool that blends cutting-edge technology with real-world usability.

I worked in the RadMadLab at UC Berkeley to research the impact of large-scale language models on neurodivergent students in educational environments. My focus is on the technical overlap between linguistic analysis and anti-plagiarism applications (like GPTZero).

I helped optimized the back-end migration process for sales, CRM, and payment data. One main task was helping bring a novel HTML5 feature through its full product cycle. I started by gathering customer data via meetings and then assisted with an initiative to increase its deployment. At the end of my internship, it was a primary tool used daily by customers. In addition, I deployed a web analytics platform that used transcriptions from recordings to optimize future sales.

I assisted in the testing and manufacturing process for high-frequency circuit boards contracted for the U.S. Army's Multiple Integrated Laser Engagement System (M.I.L.E.S). Using RF analysis software, I was in charge of verifying signal integrity and confirming accuracy with electronic schematics issued by Lockheed Martin.

I led the technical development for a JPL student engineering team tasked with designing and launching an Arduino-based rocket. I developed onboard software to integrate the rocket's real-time metrics onto a digital display, and implemented algorithms for flight telemetry that involved rigorous testing.

I worked as an automotive mechanic in La Canada, California.

Counselor for 50 middle school students introducing foundational STEM concepts through hands-on activities and individualized mentorship.