Darrin Bright

Integrated Master of Technology in Software Engineering
Vellore Institute of Technology, Vellore

My research focuses on medical image analysis, with an emphasis on computational methods for segmentation, detection and 3D biomedical image analysis. I am interested in developing methods that address challenging problems across diverse biomedical imaging applications.

Core Interests:

  •   Medical Image Analysis
  •   3D Vision
  •   Label-Efficient Learning

Education

Vellore Institute of Technology - Vellore, India

Integrated Master of Technology in Software Engineering
School of Computer Science Engineering and Information Systems (SCORE)
Courses: Artificial Intelligence, Machine Learning, Natural Language Processing, Data Mining, Computer Networks, Data Structures and Algorithms, Database Management Systems, Operating Systems, Python Programming, Object Oriented Programming.
CGPA: 8.81/10
September 2022 - May 2027
vit

Publications

  1. PortionNet: Distilling 3D Geometric Knowledge for Food Nutrition Estimation
    Darrin Bright, Rakshith Raj, Kanchan Keisham
    Journal of Computational Vision and Imaging Systems, Vol. 11, No. 1 (Proceedings of CVIS 2025)
    [Paper] [Code]
  2. A Lightweight Histology-Aware Learning Framework with Spectral Channel Recalibration and Prototype-Graph Refinement for Oral Squamous Cell Carcinoma Detection
    Darrin Bright, Bharanidharan N, Vinoth Kumar Venkatesan
    Results in Engineering (Elsevier), Under Review

Research Experience

Nanyang Technological University - Singapore (Onsite)

Research Intern
Advisor: Dr. Si Yong Yeo
  • Developing an end-to-end semi-supervised coronary vessel segmentation framework that treats segmentation as bidirectional correction, with complementary branches to recover missed vessels and remove false detections.
  • Achieved preliminary Dice improvements of 2.34 pp on the XCA134 dataset and 1.63 pp on the XCAD dataset over the state-of-the-art baseline using only 10% labeled data.
    • July 2026 - Present
    NTU Logo

    Simon Fraser University - Vancouver, Canada (Remote)

    Research Intern
    Advisor: Dr. Ghassan Hamarneh
  • Developed the first deep learning method for direct detection of mitochondria-ER contact sites in 3D super-resolution microscopy (STED and SIM).
  • Generated parametric 3D organelle geometries and rendered them through physics-based simulators to produce realistic training data, labeled with established contact-site criteria, overcoming the lack of ground truth in super-resolution microscopy.
  • Validated on real microscopy against MCS-DETECT, correlative EM ground truth and known cell-line contact differences, showing strong spatial agreement with all three.
    • December 2025 - Present
    SFU Logo

    Indian Institute of Technology, Kharagpur - Kharagpur, India (Remote)

    Research Intern
    Advisor: Dr. Debanjan Das
  • Developed GraphUNet, representing beat-level ECG/PPG morphology as graphs within a U-Net architecture for cuffless blood-pressure estimation, enabling explicit modeling of physiological relationships across heartbeats.
  • Evaluated on VitalDB and PulseDB against published baselines, achieving 4.97 mmHg SBP MAE and 2.77 mmHg DBP MAE under subject-disjoint, calibrated evaluation, demonstrating competitive cuffless BP estimation performance.
    • May 2026 - July 2026
    IIT Kharagpur Logo

    Indian Institute of Technology, Hyderabad - Hyderabad, India (Remote)

    Research Intern
    Advisor: Dr. C. Krishna Mohan
  • Built a federated continual learning framework integrating a global synthetic-data anchor with V-Information-guided selective re-distillation to preserve prior-task knowledge without sharing patient data or maintaining a conventional replay buffer.
  • Evaluated across five MedMNIST modalities, improving average final accuracy by 32.12 pp and reducing average forgetting by 20.26 pp over FedCIL.
    • January 2026 - June 2026
    IIT Hyderabad Logo

    Vellore Institute of Technology - Vellore, India (Onsite)

    Undergraduate Researcher
    SPHINX-Net: Histology-Aware Framework for Oral Cancer Detection
    Advisor: Dr. Bharanidharan N
  • Developed SPHINX-Net, a lightweight hybrid CNN–Transformer framework for patch-level oral squamous cell carcinoma detection, combining histology-aware multi-view contrastive pretraining, spectral channel recalibration, and mini-batch prototype-graph refinement.
  • Achieved 96.30% patch-level accuracy with a 1.75M-parameter model at 1,034 FPS.
    • PortionNet: Distilling 3D Geometric Knowledge for Food Nutrition Estimation
    Advisor: Dr. Kanchan Keisham
  • Developed PortionNet, a cross-modal knowledge distillation framework that transfers 3D geometric knowledge from point clouds to an RGB-only model through a lightweight RGB-to-Geometry Adapter, eliminating depth-sensor requirements at inference.
  • Achieved SOTA on MetaFood3D, with 17.43% volume MAPE and 15.36% energy MAPE, reducing errors by 57.9% and 77.4% relative to the previous best; further achieved 12.17% energy MAPE on the SimpleFood45 dataset.
    • July 2025 – June 2026
    VIT Logo

    Indian Institute of Technology, Jodhpur - Jodhpur, India (Onsite)

    Research Intern
    Advisor: Dr. Deepak Mishra
    Self-Supervised Medical Image Segmentation with Vector Quantization
  • Developed a hierarchical vector quantization framework that addresses limited labeled data in medical imaging by learning discrete representations from unlabeled ISIC 2018 images and using them as structured regularizers during segmentation fine-tuning on the PH2 dataset.
  • Designed three-stage training pipeline integrating VQ-VAE, SimCLR contrastive learning and segmentation fine-tuning, with multi-layer quantization capturing features across semantic scales.
  • Achieved 79.76% Dice, a +1.38 pp consistent improvement over baseline without VQ across multiple runs.
    • Vector Quantization for Semantic Segmentation
  • Conducted 13-experiment ablation study on Pascal VOC 2012 dataset investigating vector quantization integration with Swin Transformer encoders, testing codebook configurations, fusion mechanisms and initialization strategies.
  • Achieved 74.61% Dice score with optimized vector quantization configuration on Pascal VOC 2012, demonstrating +7.34 pp improvement over baseline CNN and +3.91 pp over vanilla Swin Transformer, validating VQ effectiveness for semantic segmentation tasks.
    • May 2025 - July 2025
    IIT Jodhpur Logo

    Professional Experience

    Astute - Vellore, India (Onsite)

    Project Manager and Technical Lead, Intern (Oct 2024 - Apr 2025)
  • Architected SEO AgentFlow, a multi-agent AI framework with six specialized agents for keyword extraction, on/off-page optimization, technical SEO, and website analysis.
  • Led development of AI-driven automation workflows integrating RAG-based content generation, web scraping, and automated content and advertisement generation.
    • GenAI/ML Engineer, Intern (Jul 2024 - Oct 2024)
  • Built a data-driven keyword research pipeline integrating SERP API, Google Trends, and DataForSEO with multi-dimensional ranking for competitive analysis.
  • Developed automated advertisement generation using Stable Diffusion for poster design and Gemini for tagline generation and placement.
    • July 2024 - April 2025
    Astute Logo

    Vicuna Kouture - Vellore, India

    AI Specialist - Intern
  • Experimented with various text-to-image diffusion models to generate unique t-shirt designs, achieving optimal results with Stable Diffusion.
  • Performed extensive prompt engineering to refine and identify the most effective prompts for generating visually appealing outputs.
  • Collaborated with the back-end team to integrate the AI model using FASTAPI, ensuring smooth deployment.
    • July 2024 - August 2024
    Vicuna Kouture Logo

    IEEE Robotics and Automation Society - Vellore, India

    Chairperson
  • Led a chapter of over 500 students, organizing hackathons, workshops, and tech talks to inspire learning in robotics and automation.
  • Mentored a team of juniors by conducting sessions on ML and Generative AI frameworks like LangChain to enhance their skills and knowledge in the field of AI and automation.
    • December 2024 - December 2025
    RAS Logo

    Skillset


    python c++ python python c++ c++ c++

    c++ c++ c++ c++ c++ c++

    c++ c++ c++ C html css js

    Notable Achievements

    • Winners of Project 2039 Hackathon Vellore Institute of Technology, 2024
    • Winners of Alphaforge Ideathon Vellore Institute of Technology, 2024
    • Runners up of Biomimicry Innovation Challenge Vellore Institute of Technology, 2024
    • Runners up of Startup Demo Day Vellore Institute of Technology, 2025
    • Second Runners up of HackWar Hackathon Vellore Institute of Technology, 2025

    Extra-curricular

    National Service Scheme, NGO - Vellore, India

    Volunteer
    Engaged with students at government schools and conducted sessions on career opportunities
    orioles

    Becoming I Foundation, NGO - Vellore, India

    Volunteer
    Participated in outreach programs, visiting government schools to teach foundational science, math, and english
    orioles

    V-Care Club - Chennai, India

    Prefectorial Body Member
    Participated in club activities such as growing plants and organizing newspaper recycling initiatives
    orioles

    Contact me

    Email ID: darrin.bright2022@vitstudent.ac.in
    My inbox is always open; glad to connect, discuss and collaborate. Could relate to work, volunteering or a casual talk!
    Feel free to send me an email ; will get back ASAP!

    Copyrights © Darrin Bright / Last update September 2026