Software + Intelligent Systems

FIRST PLACE — RECURSIVE HACKATHON 2026

Hi, I’m
Parijat.

I build intelligent systems that interact with the real world.

Explore My Work Get In Touch

parijatmajumdar.xyz

VISION / WORLD MODEL
Autonomous navigation schematic A perspective terrain grid with a segmented traversable region, two obstacles and a cyan planned trajectory from the vehicle to a goal. Schematic only, not live data. OBSTACLE TRAVERSABLE REGION PLANNED PATH UGV / POSE CAMERA / INPUT
  • AI / ML
  • Computer Vision
  • Autonomous Robotics
  • Software Development

Featured projects

From perception to possibility.

Autonomous systems. Imaging research. Ideas built to meet the physical world.

Autonomous robotics / Computer vision

ANVESHAK

Navigation beyond GPS.

A camera-based autonomous navigation system for unmanned ground vehicles, designed to explore navigation in GPS-denied environments.

  • Python
  • OpenCV
  • Semantic Segmentation
  • Visual Odometry
  • Path Planning
  1. PERCEIVE
  2. LOCALIZE
  3. MAP
  4. NAVIGATE

First Place — Recursive Hackathon 2026

Team Fine Tune · Organized by GNIT Kolkata ACM Student Chapter

View on GitHub
OFF-ROAD / SEMANTIC MAP
Autonomous navigation schematic A perspective terrain grid with a segmented traversable region, two obstacles and a cyan planned trajectory from the vehicle to a goal. Schematic only, not live data. OBSTACLE TRAVERSABLE REGION PLANNED PATH UGV / POSE CAMERA / INPUT

Medical imaging / Research

MedVision AI

Seeing beyond the image.

An AI-based medical imaging research project exploring X-rays, CT scans, and other imaging modalities.

  • Python
  • Machine Learning
  • Medical Imaging

Research-focused. No clinical validation claimed.

GitHub · Repository not supplied

X-RAY / SCHEMATIC
ILLUSTRATIVE ONLY
CT / SCHEMATIC
ILLUSTRATIVE ONLY

X-RAYS · CT SCANS · OTHER IMAGING MODALITIES

The lab / Forthcoming

More Experiments

A space for future builds, research notes and engineering explorations. New work will appear here.

Engineering pipeline

A system, not a single model.

From a camera frame to a planned trajectory: four connected stages that shape my approach to autonomous navigation.

  1. PERCEIVE

    Interpret the scene. Separate traversable terrain from the surrounding environment.

    Camera + Semantic segmentation

  2. LOCALIZE

    Estimate motion from visual cues. Build an understanding of the vehicle's pose.

    Features + Visual odometry

  3. MAP

    Connect observations into a spatial model of the world around the vehicle.

    Observations + World model

  4. NAVIGATE

    Plan a route through the mapped terrain. Turn perception into directed movement.

    World model + Path planning

CONCEPTUAL WORKFLOW / NO LIVE TELEMETRY

Skills & toolkit

Tools for intelligent systems.

Languages, vision methods and development tools that support my engineering and research focus.

Programming

  • Python
  • C
  • Kotlin

AI & Computer Vision

  • Machine Learning
  • OpenCV
  • Semantic Segmentation
  • Image Processing

Robotics & Algorithms

  • Visual Odometry
  • Path Planning
  • A* Search
  • Autonomous Navigation

Development

  • Git
  • GitHub
  • Android Development

About

Code that connects to the world.

Parijat Majumdar

parijatmajumdar.xyz

I’m Parijat Majumdar, a developer interested in building practical systems at the intersection of software, AI, and robotics.

I enjoy experimenting with ideas and turning concepts into working projects. From camera-based autonomous navigation to AI-based medical imaging research, I explore how software can interpret complex visual environments and turn that understanding into useful behavior.

Achievement

First Place, Recursive Hackathon 2026

Team Fine Tune · Organized by GNIT Kolkata ACM Student Chapter

Contact

Let’s build
what’s next.

For conversations about intelligent systems, computer vision and engineering.

  • GitHubProfile URL placeholder NOT SUPPLIED
  • LinkedInProfile URL placeholder NOT SUPPLIED
  • EmailEmail address placeholder NOT SUPPLIED
  • Contact details are placeholders. Add yours in app.py.