Dhruvilsinh Chauhan

• Collaborating in a 3-member cross-functional team to enhance humanoid-robot functionality by fine-tuning LLaMA 3.1 and Stable Diffusion 3.5, improving emotion recognition and increasing text-to-image generation relevance by 22% for human-robot interaction tasks. • Integrated custom neural layers into 1000+ lines of Stable Diffusion 3.5 code, resolving tensor shape inconsistencies and fine-tuning the model using the LAION-400M dataset, resulting in a 22% improvement in image realism and diversity. • Working on a speech-to-speech full duplex system like Moshi, enabling natural voice-based interaction within 200ms latency, with early testing showing 40% improvement in response timing and clarity compared to baseline models.

• Mentored 40 students in the 'Artificial Intelligence course', providing constructive feedback and leading tutoring sessions. • Enhanced academic performance and understanding of key AI concepts for students. • Developed strong communication and leadership skills through teaching and mentoring responsibilities.

• Provided detailed feedback and facilitated tutoring sessions for 35 master's students in 'Artificial Intelligence and Big Data course' at San Diego State University. • Strengthened students expertise in data analytics, machine learning, and big data technologies to enhance academic success.

• Engineered a computer vision product using remote Photoplethysmography (rPPG) technology to analyze facial data and predict heart rate, heart rate variability, stress levels, and oxygen saturation in real-time. • Developed a Flask API and containerized live video streaming platform on AWS using EC2, Route 53, and Elastic Load Balancer. • Secured 80% accuracy while boosting user engagement in health monitoring applications.

• Designed and implemented a dynamic web application for real-time notice posting, editing, and access, featuring timestamped updates for administrators and users using HTML, CSS, JavaScript, MySQL, and Django framework.

• Devised a breast cancer detection model during an internship, using machine learning algorithms to classify breast cancer types with high accuracy. Leveraged Python and advanced data analysis techniques to optimize model performance. • Conducted Exploratory Data Analysis (EDA) using Python libraries such as Pandas and NumPy, and trained machine learning models using KNN classification and Logistic Regression. • Achieved a model accuracy of 92.98%.

◾ I finished two tracks as a part of the 30 days of the Google Cloud program. Data Science and Machine Learning Track and Cloud Engineering Track. ◾ After the program was successfully completed, I received appreciation gifts from Google. ◾ Performed 12 quests in total with hands on labs. ✔️ Track 1: Cloud Engineering Track 1. Skill badge: Getting Started: Create and Manage Cloud Resources. 2. Skill badge: Perform Foundational Infrastructure Tasks in Google Cloud. 3. Skill badge: Setup and Configure a Cloud Environment in Google Cloud. 4. Skill badge: Deploy and Manage Cloud Environments with Google Cloud. 5. Skill badge: Build and Secure Networks in Google Cloud. 6. Skill badge: Deploy to Kubernetes in Google Cloud. ✔️ Track 2: Data Science and Machine Learning Track 1. Skill Badge: Getting Started: Create and Manage Cloud Resources 2. Skill Badge: Perform Foundation Data, ML, and AI Tasks in Google Cloud. 3. Skill Badge: Insights from Data with BigQuery. 4. Skill Badge: Engineer Data in Google Cloud. 5. Skill Badge: Integrate with Machine Learning APIs. 6. Skill Badge: Explore Machine Learning Models with Explainable AI.

Coursework: CS 514: Database Theory and Implementation CS 549: Machine Learning CS 660: Algorithm Analysis and Design CS 601: Graduate Seminar (Research) CS 576: Computer Networks and Distributed Systems CS 577: Principles and Techniques of Data Science CS 659: Visual Perception and Learning (Applied Computer Vision) CS 648: Advanced Topics in Web and Mobile Applications (Modern Web Development Frameworks) CS 653: Data Mining CS 654: Reinforcement Learning CS 649: Big Data Tools and Methods