Skip to content
View sayon-mitra024's full-sized avatar

Block or report sayon-mitra024

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sayon-mitra024/README.md

typing banner




Overview

I am an AI/ML researcher and research software engineer working at the intersection of computer vision, healthcare AI, and biomedical computing, with a supporting focus on robotics, embedded intelligence, and space science. My work runs the full research lifecycle — dataset curation, experiment design, manuscript preparation, and the software infrastructure that keeps it reproducible.

I hold a published patent on an AI-integrated blood group testing device, have released a curated biomedical imaging dataset and a research productivity framework as open-source software, and am currently preparing a confidence-aware classification pipeline for Scopus Q1 submission. Alongside this, I founded Sayon Institute of Innovation (SII), building learning and research-management infrastructure for student researchers.

Currently: B.Tech Computer Science, Chandigarh University.



Research Focus

Core Domains

  • Computer Vision & Transfer Learning
  • Healthcare AI & Biomedical Imaging
  • Confidence-Calibrated Classification
  • Bioinformatics
  • Research Software Engineering

Supporting Interests

  • Robotics & Intelligent Systems
  • Embedded Systems
  • Forensic & Security Biometrics
  • Space Science
  • Sustainable Technologies

Research Vision

I build AI systems where the reported confidence means something — models that know when to defer to a human instead of forcing a guess. My current work applies this directly to low-cost diagnostic screening, where a wrong automated call carries real clinical cost. The same principle — instrumented, reproducible, honestly-calibrated research — carries into everything from a biometric authentication pipeline to a lunar subsurface mapping model. 🚀

Featured Research

🩸 Confidence-Aware Blood Agglutination Classifier

In preparation — Scopus Q1

Frozen CNN backbone → calibrated tabular classifiers → asymmetric clinical review policy. Macro-F1 0.7411 at threshold 0.340, reliability asymmetry replicated across 5 seeds. Every reported figure is traceable to a source notebook cell.

🧬 AI Blood Group Detection & Analytics

AI_Blood_Group_Detection

MobileNetV2 feature extraction + AutoGluon tabular prediction for mobile-optimized ABO/Rh(D) screening. Concept covered by a published patent for a biodegradable dual-mode testing device.

🧪 Blood Agglutination Curated Dataset (V1)

Zenodo · 10.5281/zenodo.21384197

Quality-controlled, binary-labeled image dataset published to support reproducible research in low-cost hematology imaging.

⚙️ SayonLab — Research Productivity Framework

SayonLab

Open-source Python framework for research automation, experiment tracking, and reproducibility — extending the SII research ecosystem into public tooling.

🔍 Forensic Gait Biometric Authentication

cu-gait-biometric · Live demo

Gait Energy Image (GEI) methodology for person authentication from walking video, deployed as a full Flask application with public inference.

🛰️ IEEE CIS — Student Branch Platform

cis-ieee-cu.vercel.app

Next.js / TypeScript / Firebase platform with role-based access control, an authenticated committee portal, and automated dual-logo certificate generation.

Robotics: Aarvi — autonomous indoor delivery robot  ·  Airo — sanitation & humidification robot, First Prize robotics innovation

Publications & IP

Title Venue Domain Link
Automated Blood Group Detection Using Computer Vision and Transfer Learning on Mobile-Optimized Deep Neural Networks Zenodo Research Repository AI · ML · CV · Healthcare 10.5281/zenodo.20224963
Blood Agglutination Curated Dataset (V1) [Data set] Zenodo Biomedical Dataset · CV 10.5281/zenodo.21384197
SayonLab: A Scientific Research Productivity Framework v0.1.0 [Software] GitHub Open-Source Scientific Computing SayonLab

PatentDual-Mode Blood Group Testing Device Featuring Biodegradable Design and AI Analytics · App. No. 202511125188 · Filed 11 Dec 2025 · Published

GitHub Analytics

Selected Achievements

  • 🏆 First Runner-Up — CSIR-IITR Innovation Award
  • 🥇 First Prize — Aarvi & Airo Robotics Innovation
  • 🥉 Third Prize — Inter-School Website Development Competition

Let's Connect

ORCID LinkedIn Website Zenodo


Building reproducible AI for problems where being wrong has a cost.

Popular repositories Loading

  1. SayonLab SayonLab Public

    A Scientific Research Productivity Framework for publication-ready visualization, reproducible workflows, and scientific computing in Python.

    Python 3

  2. Attendance-IQ Attendance-IQ Public

    A Modern Offline Smart Attendance Predictor with Calendar-Based Attendance Tracking, Subject-Wise Analytics, Recovery Date Prediction, and Interactive Dashboard.

    HTML 2

  3. Student-Management-System Student-Management-System Public

    Python Tkinter-based Student Management System with SQLite database. Add, update, delete, and search student records, track monthly payments, export CSV/PDF reports, and print PDFs. User-friendly G…

    Python 1

  4. sayon-mitra024 sayon-mitra024 Public

    1

  5. Sehat-Bandhu Sehat-Bandhu Public

    Sehat Bandhu : An AI-powered, multilingual health companion for the Smart India Hackathon (SIH 2025). Provides reliable guidance, hospital booking links, and voice-to-text/text-to-speech functional…

    Python 1

  6. ChikitsaMitra-AI-Powered-Preventive-Assisted-Healthcare-Ecosystem ChikitsaMitra-AI-Powered-Preventive-Assisted-Healthcare-Ecosystem Public

    ChikitsaMitra is a comprehensive, multi-modal AI companion designed to bridge critical gaps in the Indian healthcare landscape. We address issues of access, misinformation, language barriers, and f…

    Jupyter Notebook 1