Software Developer Data Science

Kargi Chauhan

+1-520-***-**** # ****************@*****.*** ï Kargi-chauhan § Github

Work Experience

•

3UM Remote, USA

ML Engineer - LLMs April 2024 - Present

Fine-tuned Large Language Models using PyTorch and JAX, achieving up to a 25% improvement in model accuracy and efficiency.

Optimized fine-tuning and inference with advanced techniques (KV cache, reduced precision, Multi Query Attention, Rotary Embeddings) and integrate into cloud-based products, enhancing performance of distributed NLP system.

•

NASA JPL - SpaceTrex Arizona, USA

ML Research Scientist - ML, NLP Feb 2024 - May 2024

Designed advanced attitude estimation and lighting systems for a CubeSat satellite, achieving precise pose estimation using singular vision sensors.

Trained multi-modal neural networks and managed hardware design, test, and validation, achieving 97% accuracy on synthetically developed real-time data sets comprising 100,000 million records.

Successfully managed large-scale data processing comparable to AWS infrastructure, and contributed to the NASA Johnson pose estimation challenge.

•

Mining Engineering Arizona, USA

Data Science Researcher - ML,VR Feb 2023 - April 2024

Designed and implemented a realistic virtual replica of the St.Xavier Mine in Arizona, incorporating advanced technological features and expert collaborations.

Optimized algorithm efficiency for A* and Dijkstra pathfinding, cutting down execution times by 23%, which significantly improved navigational latency and responsiveness in Mine simulation.

•

Tech Core Arizona, USA

Software Developer Intern June 2022 - Aug 2022

Delivered RESTful Web APIs using Node.js and utilized Microservices architecture, employing an API Gateway to manage and coordinate the overall architecture.

Architected and deployed GraphQL APIs using Apollo server, enhanced MongoDB and Elasticsearch pipelines through indexing and query optimization, resulting in 50% faster data retriveal and 20% less server load.

•

GirlScript Summer of Code Remote, USA

Software Developer Intern Jun 2021 - Aug 2021

Developed features for a transaction web application as a full-stack developer, using React, Redux, Angular, and MySQL. Reduced ticket submission form loading time by 40% through code optimization.

Developed using Node.js, Knex.js, and Bookshelf.js to optimize and streamline ticket processing operation resulting in a remarkable 80% decrease in the time required to complete the transaction. Skills Summary

• Machine Learning: PyTorch, TensorFlow, Scikit-Learn, Apache Spark, Hadoop, Pandas, NumPy, Matplotlib, Seaborn

• Languages & Frameworks: Python, C++, C#, PHP, JavaScript, ReactJS, HTML, CSS, Swift, Flask, MongoDB, MySQL

• Software Development: Git, Docker, Kubernetes, AWS(Lambda, EC2), Unity3D Education

•

University Of Arizona Arizona, USA

BS in Information and Data Science, Minor in Game Design and Development 2020 - 2024 Relevant Courses: Software Development, Data Structures and Algorithms, Machine Learning GPA: 3.9 / 4.0 Selected Publications

• Kargi Chauhan, Vishnu Pendyala, “Large Language Models and XAI ” (Under Review)

• Kargi Chauhan, Athip Thirupathi Raj, Jekan Thangavelautham, “Enabling Deep Space Using Inspectors Accompanying Small Spacecraft System of System Architecture”, Interplanetary Small Satellite Conference, NASA JPL 2024. Paper

• Kargi Chauhan, Angelina Anani, Sefiu Adewuyi, “From Mines to Minds: Exploring Immersive Learning’s Influence in Mining Engineering Education”, UR Inspiration a Undergraduate Research conference Paper

• Hannah D Budinoff, Andrew Wessman, Kargi Chauhan, “Using online learning modules to improve students’ use of technical standards in additive manufacturing courses and projects”, ASEE 2023. Paper Academic Projects

• Pose Bowl: Spacecraft Detection and Pose Estimation Challenge: Developed computer vision algorithm for spacecraft camera position estimation improving NASA’s R5 inspection accuracy by 97%, enhancing mission safety in space. Code

• Metropolis - Hastings MCMC Inference of 3D Line: Implemented Metropolis-Hastings MCMC to estimate 3D line parameters from noisy 2D images, generating samples, finding MAP estimates, and using dual-camera data for 3D modeling in autonomous systems. Code

Contact this candidate