Accomplishments:

1. Successfully compiled and optimized a custom designed VITS model architecture on Amazon EC2 Inf2 instance, reducing inference time from 4260.83 ms to 90.46 ms, resulting in 46.1 times more throughput.
2. Rigorously analyzed the Stable Diffusion XL Turbo model and presented the model architecture, training process, inference process, and benchmarks.
3. Researched CNN and Transformer model architectures to improve my understanding, allowing me to better analyze and optimize larger DNNs.

During my internship at Rebellions, I thoroughly analyzed various machine learning frameworks and model architectures, benefiting from invaluable feedback and mentorship from seasoned industry practitioners. The internship enabled me to enhance my comprehension of AI acceleration in both hardware and software domains. I was able to make a positive impact on the company by conducting research, giving presentations, and implementing text-to-speech model architectures within specific compilation processes.

Developer for CUApts, a Cornell Housing Website to help students find off-campus housing.
Accomplishments:

• Agile Development, Version Control and Code Reviews: Led Agile development processes, completing weekly sprint tasks and reviewing pull requests, contributing to a 60% reduction in user bug reports.
• Cross-Team Collaboration: Worked with designers to improve UI/UX by making elements more interactive with scroll and hover styling using React and TypeScript, decreasing the site’s bounce rate by 15%.
• Improved Back-End Efficiency: Implemented and rewrote RESTful APIs using Express.js to retrieve apartment information more efficiently from Firebase Cloud Firestore, reducing API response time by 14%.
• Improved User Retention: Implemented an email notification system, increasing user return rate by 45%.

My Pull Requests (Sprint Tasks):

1. Add price and bedroom attributes to Firestore database script.
2. Implemented new design, animations, and functionality for dropdown menus on the website.
3. Sorting functionality for Apartments, allowing you to find apartments based on price, rating, number of reviews, and more.
4. Designed and Implemented a Review Email System, notifying users when their reviews have been approved to thank users for their contributions to the platform. Designed the email's html and ensured compatibility across email platforms.
5. Fixed UI inconsistencies across the website by communicating with the team's designers
6. Revamped FAQ Page

Reviewed Pull Requests

1. Landlord Redirect Button
2. Google Maps Implementation
3. Review Modal Redesign
4. Apartment Images
5. Contact Us Modal