Mishka Jethwani

A motivated CS student at UCSD with experience in Software Engineering and Machine Learning

Experiences

SWE Intern @ Reydeo LLC


Developed reusable React components and pages using Next.js, enhancing the UI/UX of a web application while learning the basics of server-side rendering (SSR) and static site generation (SSG).

Collaborated with a small team to integrate APIs, manage application state with React hooks, and debug front-end issues, gaining hands-on experience in troubleshooting and problem-solving.

Assisted in optimizing page load times by implementing code splitting and lazy loading, contributing to an improved user experience and faster application performance.

Research Scholar and Summer Intern @ UCSD


Got 2 scholarships for Implementing advanced Natural Language Processing techniques using LangChain and Retrieval Augmented Generation (RAG) to enhance decision-making capabilities in Autonomous Vehicle systems

Developing an iterative framework with self-feedback loops for LLMs, optimizing their performance for vehicle navigation.

Developed prompt tuning techniques including few-shot learning and chain-of-thought for improving code generation of CodeGen 350M, by incorporating CS education theory achieving from 12.8% to 27.8% on the Human-Eval datase; published a paper for same in SIGCSE with Professor Jishen Zhao

Software Development Intern @ Parfait Consultancy Firm


Built an end-to-end automated data extraction pipeline and leveraged computer vision techniques using OpenCV and OCR to analyse data from 2000+ scanned documents with over 92% accuracy on average.

Leveraged data analysis techniques using Pandas and NumPy on large-scale datasets and applied k-means clustering for customer segmentation to create insightful data visualizations with Matplotlib, reducing manual overhead by 20%

Developed a REST API using Flask to serve the document extraction pipeline, processing 100+ requests per minute.

Vice President Operations @ CSES OpenSource


Spearheaded 16 developers and 8 software leads to develop TritonScript, a MERN stack (MongoDB, Express.js, React.js, Node.js) note-sharing application, with Firebase and contribute to activist.org (Vue.js, Nuxt, PostgreSQL, Django)

Architected scalable backend systems implementing RESTful APIs and Optimized database schemas.

Implemented CI/CD pipelines using GitHub Actions. Set up automated testing with Jest and Puperteer, achieving over 90% code coverage. Used Docker for containerization and led Agile development processes.