4+ years building full-stack applied AI products

Full-stack applied AI engineer.

I am Mayur Karmakar, building production web apps and document intelligence workflows with React, Next.js, FastAPI, PostgreSQL, and Azure AI, where accuracy, traceability, and usable review matter.

Download CV Start a conversation

Capabilities

Skills

I use this stack to build applied AI products end to end: document ingestion, OCR, RAG retrieval, review dashboards, typed APIs, and responsive interfaces that make AI output inspectable.

6 tools

Applied AI workflows

Structured LLM pipelines, document extraction, RAG, and review experiences with source-level traceability.

AOAzure OpenAIGGPT-4.1ACAzure Content UnderstandingOOCRAAAzure AI SearchRRAG

6 tools

Full-stack architecture

Typed product surfaces, API boundaries, persistence, and deployment-ready backend services.

Next.js

ReactFFastAPI

Node.jsPPostgreSQLRAREST APIs

6 tools

Frontend systems

Component-driven development, state management, and performant interfaces for complex workflows.

Redux

Zustand

TanStack Query

Tailwind CSS

Shadcn

Framer Motion

8 tools

Realtime and integrations

Live exam monitoring, communication flows, payments, auth, and third-party collaboration tools.

WebRTC

Socket.IOOOpenViduJJitsiZZoomCClerkSStripePPayPal

8 tools

Cloud and delivery

CI/CD, containers, Azure hosting, and database platforms for shipping full-stack AI applications.

DDockerGAGitHub ActionsAAAzure App ServiceASAzure Static Web AppsVVercelSSupabaseNNeonTTurso

6 tools

Data and ORMs

Practical data modeling, persistence, and review screens that keep AI output auditable.

Drizzle ORM

Prisma ORM

tRPC

Python

Django REST

Pandas

Selected work

Experience

My work now spans applied AI, regulatory intelligence, EdTech proctoring, and healthcare analytics: full-stack systems where the interface has to make model output, realtime events, and protected data easy to inspect.

Applied AI document intelligence

AI Review Platform

zipp.ai ~ Oct 2025 - May 2026

Structured review

A full-stack, AI-assisted document-review product that turns complex source material into structured, reviewer-ready insight.

Delivered

Designed and delivered the end-to-end product across Next.js interfaces, FastAPI services, PostgreSQL, browser automation, and cloud AI integrations.
Built resilient ingestion and structured-processing workflows with validation, retry handling, persistence, and traceable outputs.
Delivered analytics surfaces with multidimensional exploration, comparative views, trend analysis, and reviewer-focused workflows.
Connected AI-generated outputs to source context, enabling faster inspection and higher-confidence human review.
Prepared application services for repeatable cloud delivery with containers, CI/CD, and managed AI integrations.

Applied AI document intelligence

Document Review Workflow

zipp.ai ~ Oct 2025 - May 2026

OCR and review

An AI-assisted document-review workflow for turning complex operational records into structured, traceable reviewer insight.

Delivered

Built OCR-backed document-processing flows that converted complex PDFs into structured, review-ready information.
Implemented validation and quality-check workflows that surfaced incomplete or inconsistent information for human follow-up.
Linked structured findings to source context, helping reviewers verify outputs quickly and confidently.
Delivered auditable review experiences and workflow support that improved traceability across document versions.

AI learning workflow

Content2Quiz

Bsecure - 2023 to 2025

Question banks

An AI-powered quiz generator for transforming text and uploaded files into structured question banks.

Delivered

Developed a Next.js, React, Drizzle ORM, and Tailwind application for quiz creation, review, and exam workflows.
Built OpenAI GPT flows that converted typed text and uploaded files into structured question banks.
Implemented RBAC, protected flows, exam scheduling, and automatic submission behavior.
Integrated PayPal and Stripe payments while keeping the content-generation UI responsive across devices.

Online proctoring

Bsecure Pro

Bsecure - 2023 to 2025

Realtime integrity

A self-contained educational technology platform for online proctoring and exam management.

Delivered

Developed and deployed proctoring workflows using Next.js, React, Drizzle ORM, and Tailwind CSS.
Implemented live video and audio monitoring with WebRTC and OpenVidu, including screen recording and real-time violation alerts.
Built AI-based face and object detection flows to support exam integrity checks.
Integrated role-aware dashboards, live support chat, real-time messaging, and payment handling for proctoring operations.

LMS integration

Examd - Canvas

Bsecure / Wisteli - 2021 to 2025

Canvas workflows

An online proctoring and exam management platform that integrates with LMS and test engines.

Delivered

Revamped legacy exam and monitoring surfaces into clearer workflows for students, instructors, and proctors.
Implemented Canvas LMS assignment handling, configurable scheduling, countdown timers, auto-submission, and post-exam review.
Built real-time student-proctor communication using WebRTC, Socket.IO, and video/audio streaming.
Managed time-based exam windows, violation reporting, and LMS-compatible proctoring behavior.

Healthcare analytics

CarelogiQ

Wisteli - 2021 to 2023

Clinical dashboards & Mobile App

A HIPAA-compliant chronic disease management solution for value-based care teams.

Delivered

Contributed to React and React Native healthcare workflows for chronic disease management.
Improved diagnosis preprocessing and clinical data visualization for care-team dashboards.
Designed responsive analytics views for patient vitals and healthcare trend interpretation.
Resolved platform-specific issues across Android and iOS that affected patient monitoring flows.

Builds

Projects

Personal and public builds that show experimentation beyond company work: public-data products, student resource workflows, and configurable AI chat interfaces.

Public project

CowinHistory

A vaccine-slot history and trend application that used past availability data to help users understand booking patterns.

Historical vaccination-slot analysis
Availability trend estimation
Telegram alert delivery
React and Django REST workflow

React

Redux

React Router

Python

Django REST

MySQL+6

Code

Public project

Collegepapers

A centralized course resource platform for notes, previous year question papers, college details, and syllabus access.

Course-specific resource library
Student-first download flow
Centralized academic material
React and Django REST stack

React

Redux

React Router

Python

Django REST

MySQL

Code

Public project

Chatbot

An AI-powered conversational app for interacting with OpenAI GPT through a configurable web chat experience.

Model-selection workflow
Temperature controls
API-key configuration
Prompt and response layout

ReactOOpenAICUChat UI

Code Live

Foundation

Education

A computer applications background paired with production work in applied AI, React, FastAPI, regulatory workflows, EdTech, and healthcare platforms.

Degree

Bachelor of Computer Applications

Trident Academy Of Creative Technology, Bhubaneswar

August 2018 - September 2021CGPA 8.08

Proof points

Certifications

Formal learning that supports the production stack: React for modern interfaces, Python and Django for backend fluency.

Mar, 2021

React - The Complete Guide 2023

Udemy

UC-840177ec-c9f4-41f9-a39a-27d34e4af316

View certificate

Mar, 2018

Python and Django Full Stack Web Developer Bootcamp

Udemy

UC-ZPNYJ3TO

View certificate

Contact

Let's build something steady.

Send a short note about the AI workflow, product surface, or document-heavy process you want to improve. I read for context first, then respond with the next useful step.

Clear project notes help most

Timeline, users, documents, stack, and the current review bottleneck are enough to start a useful conversation.

Best fit

Applied AI applications, RAG and OCR workflows, React and Next.js interfaces, FastAPI services, and review dashboards.