layout |
---|
base |
{% include project_toc.html %}
## More Info {% include projectlist.html %}
My name is Vijay Kethana, and I'm a computer science student at UC Berkeley aspiring to work as a software engineer. I'm working towards specializing in AI and machine learning. I'm also an avid reader and language learner.
- Reading Log
-
RSS (Atom) Feed: [vkethana.com/feed.xml]({{ "/feed.xml" | prepend: site.baseurl }})
- GitHub: github.com/vkethana
- LinkedIn: linkedin.com/in/vkethana/
- Currently pursuing Bachelor of Arts in Computer Science (3.97 GPA) from UC Berkeley (2023 – ).
- Coursework: Data Structures, Efficient Algorithms, Computer Architecture, Probability & Random Processes, Discrete Math, Linear Algebra, Multivariable Calculus.
- Languages: Python (Pandas, Matplotlib, OpenCV), C, C++, Assembly, Java, Lisp, HTML/CSS, JavaScript, SQL
- Frameworks: React, Node.js, Flask, OpenMP, OpenMPI, JUnit
- Developer Tools: Git, Vim, IntelliJ IDEA, Docker
- System Administration: Debian, Arch Linux, DigitalOcean
June 2024 - Present
- Collaborating with UC Berkeley CS PhD students to enhance R2E, a framework for evaluating code generation and optimization capabilities of large language models (LLMs). Collect install and test commands from 50 Python GitHub repositories containing approximately 130 code optimization problems.
- Developing an LLM-powered agent to automate the cloning, setup, and execution of Python GitHub repositories; detecting and fixing installation errors using GPT-4 powered agentic loop.
2022-2023
- Organized Gilroy Hacks, a multi-school, student-led hackathon series for middle/high schoolers in Gilroy, CA.
- Secured a $1,481 grant by convincing sponsors and resolving logistical issues. Coordinated with judges and backers, including Wolfram Language, 1Password, and AoPS.
- Support development of 7 hackathon projects in Python, Next.js, Typescript, and Flask.
June - August 2022
- Developed Python script with OpenCV to track fast-moving helium balloons using a smartphone camera. Utilized the CSRT (Channel and Spatial Reliability Tracking) and KCF (Kernelized Correlation Filter) algorithms. Collaborated with two teammates under Dr. David Argudo, Ph.D.
- Processed balloon motion videos into CSV location data files. Applied Statsmodels to clean datasets and handle out-of-frame objects.
- Authored comprehensive, ten-page document describing how to conduct classroom physics demonstrations using the balloon tracking software.