Sayon Mitra

About

AI research aimed at systems that know their own limits.

I work across computer vision, healthcare AI, and robotics — building diagnostics, smart security frameworks, and bioinformatics systems that hold up under real-world conditions, not just benchmark conditions. I'm currently reading Computer Science and Engineering at Chandigarh University.

  • Computer vision
  • Healthcare AI
  • Applied robotics
  • Bioinformatics

How I work

Three things that don't change between projects.

Confidence is part of the output

A prediction without a sense of its own reliability is only half an answer. I build systems that flag what they are unsure of — a calibrated classifier with a review policy for the cases it shouldn’t call alone — rather than presenting every guess with equal certainty.

Every claim traces back to a run

Results earn their place in a manuscript only once they are reproducible from logged outputs — no number goes in that cannot be pointed to.

Useful before elegant

Healthcare and security systems are judged by how they behave at the edges. I optimize for dependable behavior under real-world conditions first, sophistication second.

Origin

Sayon Institute of Innovation started as workshops, not a website.

Since April 2023 I've been designing and delivering AI/ML workshops for Classes VI–XII across government and government-aided schools in rural West Bengal and Uttar Pradesh. That work grew into Sayon Institute of Innovation — an ecosystem that now includes Pariksham, a proctored learning and examination platform, and Anvesha, a research collaboration platform for student researchers. The build-out is documented on the projects page; this is just where it started.

Sayon Institute of Innovation

Elsewhere on this site

The specifics live on their own pages

Research→

Peer-reviewed and preprint work — including a confidence-aware blood-group classification pipeline on a Scopus-track submission, plus published datasets and open-source tooling.

Projects→

The systems themselves: the SII product suite (Pariksham, Anvesha), the IEEE CIS Chandigarh University platform, and the forensic biometric gate system in progress.

Patents→

A published patent for a dual-mode, biodegradable, AI-assisted blood group testing device.

Toolkit

What I build with

AI & machine learning

  • Machine learning
  • Deep learning
  • Computer vision
  • Transfer learning
  • Neural networks
  • Predictive analytics

Programming languages

  • Python
  • Java
  • C
  • JavaScript
  • TypeScript
  • HTML
  • CSS

Frameworks & libraries

  • OpenCV
  • Scikit-learn
  • NumPy
  • Pandas
  • AutoGluon
  • React.js
  • Node.js

Databases

  • MySQL
  • SQLite
  • Firestore

Tools & concepts

  • Git
  • Arduino IDE
  • Android Studio
  • Overleaf
  • Data structures
  • OOP
  • DBMS

Languages

English — Professional

Bengali — Advanced

Hindi — Intermediate

Assamese — Basic

Japanese — Basic

Start a conversation→