I am a current undergraduate at the University of California, Los Angeles, majoring in Computer Science (3.82 GPA, expected June 2027). I am currently a Software Engineer Intern at Shopify, working on Search (Index Serving), improving Catalog Search performance in C++ with Elasticsearch, approximate nearest neighbor indexing, and large-scale (1B+ item) catalog optimization. My experience spans software engineering, ML research, embedded systems, robotics, and web development.

The blend of systems, theoretical, and AI-focused coursework has equipped me with a strong foundation to tackle complex problems. I’ve completed upper-division courses in Algorithms (CS 180), Operating Systems (CS 111), and Programming Languages (CS 131), alongside applied ML in Data Mining (CS 145) and Natural Language Processing (CS 162). I’ve built systems in Digital and Computer Architecture (CS M152A, CS M151B), with foundational training in Statistics (C&EE 110) and Microeconomics (Econ 11). In Software Construction (CS 35L), I developed proficiency in debugging, testing, and collaborative Git workflows, skills I apply at LA Hacks, where I’ve built backend features supporting 6k+ applications through PostgreSQL, Supabase, and Next.js on lahacks.com and apply.lahacks.com.

I am a researcher at the Structures-Computer Interaction Laboratory, advised by Prof. M. Khalid Jawed and Postdoctoral Researcher Tuan-Anh Vu. My work includes HiddenFruits: an embedded/ML system for fruit detection with multi-modal CV on NVIDIA Jetson Orin Nano, developing Linux kernel drivers for FLIR Lepton thermal sensors and implementing SPI/I2C/UART protocols. I also contributed to Reconstruction Using the Invisible (AAAI 2025), focusing on thermal, NIR, and depth fusion for 3D Gaussian Splatting.

I previously researched 3D Vision-Language Transformer Models at the GRASP Laboratory (University of Pennsylvania), developing the Avi architecture, a 3D Vision-Language Action Model accepted into the NeurIPS 2025 Workshop on Embodied World Models. I ran 2k parallel simulations on NVIDIA HPC with Python, CUDA, and PyTorch.

I have also held software engineering roles at Leidos (Sea Archer autonomous naval vessels: Docker, Kubernetes, Ansible) and the Office of Naval Research (IED detection and defusal software with OpenCV, RGB-D imaging, and Turtlebots on NVIDIA Jetson TX2).

Research Interests

• Computer Vision: Object Detection, 3D Reconstruction
• Machine Learning: Domain Adaptation and Transfer Learning, Vision-Language Models, Reinforcement Learning

Selected Projects

	Hidden Fruit: A Multimodal Framework for Fruit Detection Harris Song Structures-Computer Lab - University of California, Los Angeles Project Page  |  Paper  |  Hidden Object Paper
	AI-Generated Text Detection with SVM and LoRA-Finetuned RoBERTa Harris Song, Marie Yang, Anish Pal, Aditya Patil Natural Language Processing Course - University of California, Los Angeles Paper
	EcoLens: Automated Trash Detection and Sorting Framework Harris Song, Oscar Chen, ‍Charles Liggins University of Pennsylvania Hackathon Source Code |  News Coverage
	Predicting DoorDash Delivery Time through Classical Machine Learning Harris Song, Abhinav Amanaganti, Ethan Maldonado, Afnan Khawaja Undergraduate Data Science Course - University of California, Los Angeles Paper

Publications

	Avi: A 3D Vision-Language Action Model Architecture generating Action from Volumetric Inference Harris Song, Long Le NeurIPS 2025 - Embodied World Models for Decision Making Workshop Project Website  |  PDF
	Reconstruction Using the Invisible: Intuition from NIR and Metadata for Enhanced 3D Gaussian Splatting Gyusam Chang, Tuan-Anh Vu, Vivek Alumootil, Harris Song, Deanna Pham, Sangpil Kim, M. Khalid Jawed AAAI 2025 - 39th AAAI Conference on Artificial Intelligence Paper
	Demonstration of Wireless Synchronization Methods in Autonomously Controlled Fleet of Drones Christopher Lai, Harris Song, Aaron Madrigal, Michael Youssef, Borick Lieng, Mohamed Hamida, Quyen Tran, Phu Ngo, Bethany Chang, Steven Dobbs, Zhen Yu International Journal of Mobile Network Design and Innovation, 2024 Paper