⌘KCommand
IST Kolkata, India
Available for internship

Abirbhab
Dasgupta

Code. Automate. Ship. Repeat.

Building in public. Open to opportunities.

View WorkGitHubLinkedIn
Scroll
Next.js
TypeScript
React
Component Library
Better Auth
Drizzle ORM
n8n Workflows
Open Source
Postgresql
Tailwind CSS
Git
LangChain
Automation Agents
Next.js
TypeScript
React
Component Library
Better Auth
Drizzle ORM
n8n Workflows
Open Source
Postgresql
Tailwind CSS
Git
LangChain
Automation Agents

Selected Work

Things I have
actually shipped.

StudioSynq
01
Next.jsTypeScriptVercel AI SDKDrizzle ORMNeonPusherUpstash Redis

StudioSynq

Problem — Developers, designers, and writers constantly switch between AI tools, losing context mid-workflow

Key decision — Chose Vercel AI SDK with multi-provider routing over a single LLM integration — lets each agent use the best-fit model for its task (Google, Groq, Mistral)

5 specialized agents · Real-time rooms via Pusher · Kanban with dnd-kit · Multi-LLM routing · Tavily web search

Live DemoGitHub
WorkForcee
02
Next.jsTypeScriptBetter AuthDrizzle ORMPostgreSQLTailwind CSS

WorkForcee

Problem — Small teams track attendance, leave, and payroll across scattered spreadsheets and chat threads — no single source of truth, no audit trail

Key decision — Chose Better Auth's admin plugin (ban/unban, setUserPassword) over building custom offboarding and password-reset logic — reused built-in session revocation instead of hand-rolling account lifecycle management

Role-based visibility (admin/employee) · Auto-computed payroll (Basic/HRA/PF/tax) · Live check-in/out status · Leave balance tracking · Admin-mediated password recovery

Live DemoGitHub
PromptBuildr
03
Next.jsTypeScriptGemini API

PromptBuildr

Problem — Raw ideas fed directly into AI models produce weak, generic outputs

Key decision — Chose localStorage-only storage over a backend — keeps it 100% private and zero-latency with no auth overhead

6 prompt modes · 4 target models · Zero server storage · Prompt explainability

Live DemoGitHub
FramixUI
04
Next.jsTypeScriptTailwindMarkdown

FramixUI

Problem — Building consistent UI components from scratch wastes development hours

Key decision — Chose a headless architecture to give consumers full styling control

12 components · Lighthouse 98 · Used across 2 live projects

Live DemoGitHub
AUth Next
05
Next.jsBetter AuthDrizzle ORMNeonTypeScript

AUth Next

Problem — Production-grade auth in Next.js is poorly documented and fragile

Key decision — Chose Better Auth over NextAuth for its type-safe API and native Drizzle integration

Google OAuth · Email auth · Deployed on Vercel

Live DemoGitHub
n8n AI Agents
06
n8nGoogle Gemini APITelegram Bot API

n8n AI Agents

Problem — Repetitive personal tasks like expense tracking and email filtering waste hours every week

Key decision — Chose n8n over custom code for its visual workflow builder — faster iteration, easier debugging

3 agents · Expense tracker · Calendar summariser · Gmail filter

GitHub

About

The person behind
the code.

I don't build things to pad a resume. I look for real problems worth solving — then I research them, sketch a blueprint, wire up the design, and ship something that actually works.

My interest sits at the intersection of Software Developement, AI automation, and product thinking. I've built UI libraries, production auth systems, and AI agents that automate real workflows — not just tutorial clones.

I'm looking for an internship where I can learn how real products are architected and shipped — not just how to write features, but how to think about systems from the ground up.

Education

Adamas University

B.Tech Computer Science & Engineering

2024 – 2028

PM Shri Kendriya Vidyalaya Birbhum

Higher Secondary — CBSE

Graduated 2024
ApproachResearch → Blueprint → Design → Ship
StackNext.js · TypeScript · React · Node.js · PostgreSQL · Python · Tailwind · Git

Writing

Thinking out loud.

All posts →

2026 · 5 min read

Loop Engineering vs. Prompt Engineering: Is This Actually a New Skill?

Loop engineering is the new buzzword in AI development — but is it a real shift or just prompt engineering with better branding? An honest breakdown.

2026 · 19 min read

From Function Calling to MCP: Building Your First Standardized AI Agent

I broke my function calling agent trying to scale it. This is what I learned rebuilding it with MCP — and why the architecture difference matters more than I expected.

2025 · 5 min read

Invoice Generator with AI

Building an AI-powered invoice generator — the problem, the approach, and what I learned.

More articles coming soon — writing about AI, automation, and shipping products.

Contact

Open to internships.
Let's talk.

I'm actively looking for Summer 2026 internship opportunities. If you're building something interesting and need someone who can ship — reach out. I typically reply within 24 hours.

Click to copy email address

GitHub@abirbhab-dasgupta
See what I'm building
LinkedIn@abirbhab
Professional profile
Twitter@Abirbhab_24
Building in public
© 2026 Abirbhab Dasgupta.Kolkata, India · IST