Available for Opportunities

Hi, I'm Abhisek

Full-Stack DeveloperBuilding AI-Native Systems

I build production systems where AI is the foundation, not a feature bolted on at the end — from a multilingual AI tutor with async worker pipelines to LangGraph agents that query live business data, deployed on AWS, Docker, and modern CI/CD.

View ProjectsDownload CV
Profile Picture of Abhisek Gupta

TECHNOLOGIES I WORK WITH

AI & Agents

LangChain
LangChain
OpenAI
OpenAI
Gemini
Gemini
FastAPI
FastAPI
Python
Python

Infrastructure

AWS
AWS
Docker
Docker
Redis
Redis
Vercel
Vercel
GitHub
GitHub

Backend

Node.js
Node.js
PostgreSQL
PostgreSQL
Drizzle ORM
Drizzle ORM
Stripe
Stripe
Clerk
Clerk

Frontend

Next.js
Next.js
React
React
TypeScript
TypeScript
Tailwind CSS
Tailwind CSS

🚀 SELECTED WORK

Recent Projects

Available for new projects
Cognora project screenshot

Flagship — AI-Native System

Cognora

A multilingual AI tutor for LMS platforms — supporting English, Hindi, and Nepali — that teaches, quizzes, and evaluates students automatically

Role: Owned the full stack end to end — product, AI pipeline, backend, frontend, infrastructure, and deployment

Problem: LMS platforms needed more than a chatbot — students needed tutoring, assessment, and reporting in their own language, without blocking the app while AI processes long-running tasks

Architecture:

  • 4-service system: Next.js frontend, FastAPI backend, dedicated worker, and Redis as the job queue
  • Async pipeline: long-running AI tasks (transcript processing, quiz generation) run on the worker so the API never blocks
  • Multilingual tutoring with streaming responses across English, Hindi, and Nepali
  • Automated evaluation and personalized performance reports delivered by email

Deployment: Containerized with Docker, deployed on AWS EC2 with a CI/CD pipeline for automated releases

Outcome: An AI-native product designed around the AI workflow from day one — not a feature bolted onto an existing app

Next.jsFastAPIDockerRedisAWS EC2CI/CDLangChain
View Project
OpScale project screenshot

AI Agent Integration

OpScale

An AI-powered wholesale operations platform — dealers query overdue payments, revenue, and sales trends in plain English

Role: Designed the data model and integrated a LangGraph AI agent into an existing operations platform

Problem: Wholesale dealers track payments manually across notebooks and spreadsheets — they needed a way to ask questions about their own business and get real answers

AI Agent:

  • LangGraph agent with 5 business intelligence tools, scoped per organization (org_id isolation)
  • FastAPI backend on Railway streams agent responses to the Next.js dashboard via SSE
  • Credit ledger (khata) system tracking deferred and partial payments per buyer
  • Write operations stay in the validated Next.js layer — the agent reads and drafts, humans confirm

Deployment: FastAPI deployed on Railway, Next.js on Vercel, PostgreSQL on Neon

Outcome: A working example of adding an AI layer to existing business logic without compromising data integrity

Next.jsTypeScriptPostgreSQLLangGraphFastAPIRailway
View Project
PayBridge project screenshot

Open Source · Published on npm

PayBridge

A production-ready, framework-agnostic TypeScript SDK unifying Nepali payment gateway integration

Role: Designed and published a TypeScript SDK from scratch — API design, documentation, and developer experience

Problem: Every developer integrating Nepali payment gateways hits the same wall — no npm package, contradictory docs, inconsistent parameter names across providers

SDK Design:

  • Published @paybridgejs/khalti on npm — a 2-function API (initiate / verify) with full type safety
  • Framework-agnostic: works identically in Next.js, Express, and vanilla Node.js
  • Turborepo + pnpm monorepo with working examples for every supported framework
  • eSewa SDK in active development, following the same unified interface

Deployment: Documentation site built with Next.js, deployed on Vercel; package published and versioned on npm

Outcome: A real install-and-ship developer tool — the Stripe-style developer experience Nepal's payment ecosystem was missing

TypeScriptnpmTurborepoNext.jsOpen Source
View Project

About My Engineering Work

I design AI-native systems end to end — architecture, backend, frontend, infrastructure, and deployment. That means treating AI as the foundation of the system from day one: async job pipelines, agent tool design, and the infrastructure that keeps them reliable in production, not just in a demo.

Beyond these three, you can find additional work and open-source contributions on my GitHub profile in the contact section below. I focus on building reliable systems that solve real problems — for real users, not just for a portfolio.

GET IN TOUCH

Let's talk about building AI-native systems

I'm actively interviewing for full-stack and AI engineering roles. Let's discuss building reliable AI pipelines, scalable APIs, and production-grade infrastructure.

Email

abhisekgupta.dev@gmail.com

WhatsApp

Active 24/7

+977 9811728625

Connect Elsewhere

Find me on these platforms and let's connect.

Download CV