Diya Jim
-
Mail code: 0102Campus: Tempe
-
Student Information
Undergraduate StudentComputer Science
Ira A Fulton Engineering
Diya Jim is a senior at ASU completing a B.S. in Computer Science with a Math minor (4.0 GPA), with over 8 years of research and industry experience in computer vision, human-computer interaction, and generative AI. Her current work includes developing vision systems at Intel Labs, a robotics internship with Intel–Boston Dynamics, and leading applied projects across academia and industry, resulting in research papers and patents in progress. As an emerging expert in computer vision, she seeks to advance new models and systems that enhance human experience.
Computer Vision, Human-Computer Interaction, Efficient AI Modeling
Speech & Brain Research Lab, ASU (Advisor: Prof. Ayoub Daliri) — HCI Research Intern | Jan 2023 – Present
- Created brain-computer interface in MATLAB for speech rehabilitation to record, analyze, & visualize EEG scans.
- Developing a 15+ module Python package for neurofeedback training steps.
- Authoring research paper analyzing EEG data from Auditory Brainstem Response to study hearing loss in children.
DAAD-RISE, University of Wuppertal (Advisor: Yannik Hann) — Generative AI Research Intern | May – Aug 2024
- As floods have increased in frequency in western Germany, current models do not contain enough historical data to make accurate and timely predictions, resulting in high levels of damage from recent floods.
- In order to solve this data scarcity issue, I generated a synthetic dataset of radar images.
- Generated images by building/training a vector-quantized variational autoencoder models in PyTorch on NVIDIA GPUs (Cassandra cluster)
- Performed preprocessing, statistical analysis, and augmentation to this dataset
-
Co-authoring research paper on augmentation strategies and improvements in data generation.
University interview on my experience :)
https://www.tmdt.uni-wuppertal.de/de/lehrstuhl/internationale-gastwissenschaftlerinnen/diya-jim/
News article where I was being accepted to this program :) https://news.asu.edu/20240514-science-and-technology-record-number-asu-students-selected-prestigious-daadrise-program
AI in Counterspeech Lab, ASU — NLP Student Researcher | Aug – Dec 2023
- Developed a Python NLP tool to detect disinformation and generate counterspeech responses on local Intel CPU.
- Applied tokenization and combined open-source deep learning models, tested on standard hate speech datasets.
Intel Labs, Intel Corp. (Advisor: Dr. Sainan Liu) — Computer Vision Research Intern | May 2025 – Present
- Improving video generation for assembly/disassembly tasks on NVIDIA GPUs via SLURM cluster, focusing on temporal consistency and part interactions.
- Developed ComfyUI workflows for controllable video generation of assembly/disassembly from 3D models, integrating segmentation (SAM 2), object detection (YOLOv11), motion tracking, and prompt tuning.
- Created datasets of AI-generated assembly/disassembly videos from 3D models in the PartNet-Mobility dataset to fine-tune foundation models.
- Researched strategies to improve model articulation accuracy and realism.
- Built AI-based realism evaluation system using Qwen2.5-VL-3B-Instruct, including multi-video ranking, first/last frame analysis, and automated filtering.
- Researching fine-tuning strategies using generated dataset to improve part articulation accuracy and realism.
- Co-authoring two research papers on dataset creation and model fine-tuning results, planned for demo on AI Playground.
Intel Corp. / Boston Dynamics (Advisor: Dr. Alon Schlar) — Robotics / Computer Vision Intern | Jan 2025 – Present
- Built AI inference pipeline for industrial automation, making a video anonymization & hazard detection agent that runs on local Intel CPUs and generates/analyzes outputs from state-of-the-art models.
- Enabled human-robot collaboration using Spot robots for inspection and automated video inference.
- Integrated video streaming & fleet management for Spot robots using CPU-only servers & robotics middleware frameworks.
- Prototyped agentic AI for autonomous robot action selection from sensor data.
-
Collected and curated proprietary spill detection datasets to improve model robustness.
Here's a quick article on how we are using Spot robots to advance automation: https://newsroom.intel.com/intel-foundry/postcard-from-intel-foundry-direct-connect-chip-roams-intel-factories
Trenser Technology Solutions — Computer Vision SWE Intern | May – July 2023
- Designed and implemented a photo digitization application in Python, ran on-edge inference on local Intel CPU.
- Built pipelines for image processing and object detection using OpenCV and pre-trained TensorFlow models.