Divyanshu Malik

• Developed Nexus, a Next.js AI research assistant for interacting with distributed systems literature via retrieval-augmented generation (RAG). • Built ingestion pipelines using LlamaIndex, Gemini embeddings, Postgres + pgvector, and implemented persistent session-based memory. • Integrated DeepSeek LLM with inline citations, real-time streaming, and contextual code generation for enhanced UX. • Delivered full-stack features in React, TailwindCSS, and Radix UI, including PDF previews, multi-document chat, and interactive code composer.

• Contributed to NIH-funded research supporting a $6M grant, building a real-time 3D surgical scene reconstruction framework using Gaussian splatting, PyTorch, and YOLOv8, enabling interactive tumor visualization during surgery. • Implemented incremental model training and real-time monitoring, optimizing PSNR, SSIM, and LPIPS metrics, improving intraoperative decision-making. • Built full-stack deployment pipelines with Flask, React.js, and Socket.IO for real-time data visualization. • Demonstrated clinical feasibility of AI-assisted surgery by integrating dynamic model adaptation to changing camera positions and tissue deformations.

Graduate teaching assistant for a Class of 70 students in ECS 193 Senior Design Project in the Computer science dept.

• Implemented Python scripts to automate data fetching and storing using API. • Implemented and updated functions in MATLAB used for analysis of the cost of invasive species. • Worked on updating an existing market model of U.S. fruit and vegetable data and collected data related to fruit and vegetable production, prices, consumption, and trade.

• Worked on building frontend features using React.js, and supported backend services with Node.js. Developed and tested RESTful APIs to ensure smooth communication between the frontend and backend, while using MongoDB for data storage. • Gained expertise in frontend technologies and became proficient in Scrum and Agile methodologies.

• Developed an Android application utilizing deep learning and computer vision to predict COVID-19 from Chest X-ray images. Implemented transfer learning on various pre-trained models (VGG16, ResNet50, MobileNetV1) to classify COVID-19 and normal pneumonia cases, achieving an accuracy of 98.4%. • Tested on a dataset of 6,213 X-ray images, leading to improved diagnostic predictions, and the findings were published in a research journal and presented at conferences, contributing to the AI-based detection of COVID-19.

• Created an Android app for performance athletes using Kotlin in Android Studio, with threading (Coroutines) for better performance and XML for UI design. • Integrated Firebase for backend, Google ads for monetization, and YoutubePlayerAPI for video support. The app hit over 15,000 downloads in its first year significantly increasing revenue for the company.

Course Work : - Human Computer Interaction - Distributed database systems - Machine Learning and Discovery. - Advanced Computer Architecture. - Design and Analysis of Algorithms. - Computer Animation. - Storytelling Using Artificial Intelligence.