V. Anbuchelvan

Open to internships and engineering roles

I build software businesses run their day on.

I work on every layer of Eesa AI, an agent platform that is live in production: the phone app, the web app, the backend and the add-ons the agent uses as tools. On the side I ship open-source Mac apps and local AI tools, and bring the odd dead computer or motorcycle back to life.

See the work GitHub
Email LinkedIn
V. Anbuchelvan

CS undergrad / SRM

V. AnbuchelvanBuilding Eesa AI
  • 1,400+commits on Eesa AI in four months
  • 9repositories across the product
  • 5add-ons the agent uses as tools
  • 168automated tests in my open-source projects

01What I do

What I build, and what I build it with

Five kinds of work, each next to the tools I reach for when I do it.

  • A / 01

    AI agents and local AI

    Tools an agent can call, workflows that wait for a person's answer, and approval before anything is written to an outside system. Local models for chat and vision.

    MCPLangGraphClaude APIAgent tool designOllamaPyTorchYOLOOpenCV

  • A / 02

    Phone apps

    The iPhone and Android app for Eesa, written in Flutter: chat, boards, and sharing a file in from any other app.

    FlutterDartSwiftReact Native

  • A / 03

    Web and backend

    Screens in the browser and the services behind them, with tests and builds that run on every push.

    PythonFastAPIFlaskNext.jsTypeScriptNode.jsPostgreSQLDockerGitHub Actions

  • A / 04

    Mac apps and systems

    Native utilities that reach where apps usually do not: the keyboard's LEDs, the trackpad's haptics, the sound the Mac is playing.

    SwiftCmacOS internalsScreenCaptureKitXCTestLinuxTrueNAS

  • A / 05

    AI-assisted development

    I build with coding agents every day, and back what they write with tests and automated builds.

    Claude CodeGoogle AntigravityGitHub Copilot

02Selected work

What I built, and why

The product I work on every day comes first. After it, the tools I made because I wanted them to exist.

Product / Live in production

Eesa.ai

A business runs its operations by asking.

Eesa connects to the systems a company already uses, builds the screens its team needs, and does the work: on the web, on iPhone and Android, and in WhatsApp.

Visit eesa.ai

Problem

A small business's day is spread over apps, spreadsheets and chat threads that do not talk to each other.

Product

One agent that reads those systems, builds a screen for each job and carries out the request, with a person's approval where it matters.

My part

The phone app, five add-ons, workflow and approval work in the backend, and the same screens on the web.

My work on Eesa, layer by layer

  • Phone app

    iPhone and Android

    Primary contributor to the app: chat, the Flow Board, and sharing a file to Eesa from any other app.

    Flutter / Dart / Swift

  • Add-ons

    Tools the agent can use

    Primary contributor to five add-ons that plug into the agent as tools and show up in chat, on the boards and on the team's phones.

    Node.js / MCP / PostgreSQL

  • Backend

    Workflows, roles, approvals

    Workflows that run on a schedule and wait for a person's answer, role-based access for add-ons, approval before the agent writes to an outside system, and the server-driven share page.

    Python / LangGraph

  • Web app

    The same screens in the browser

    The flows from the phone, built again for the web.

    Next.js / TypeScript

  • Attendance

    Staff clock in on site from their phone, and timesheets add themselves up.

  • Drive

    Team folders and sharing.

  • Invoices

    Supplier invoices shared in from a phone, and sent on once approved.

  • Calls

    102 read-only agent tools generated from Dialpad's OpenAPI spec.

  • QuickBooks

    The agent reads the accounts. Every write waits for an approver.

Open source, built because I wanted it

The SlideView library showing lecture decks grouped by subject

Mac app / Swift

SlideView

A native Mac study app. Every lecture file in one dark, offline library, with a drawing board and a map of how your material connects.

58 tests / Released as a download

Website and download
The Aether chat welcome screen

Local AI / Python

Aether

Private AI chat that runs entirely on your own machine with any Ollama model: text, images and hands-free voice.

Python / Ollama

Website
A MacBook in a dark room with its keyboard backlight lit

Mac app / Swift

Keyboard Strobe

Flashes a MacBook's keyboard backlight in time with whatever is playing, by driving the LEDs through a private Apple framework.

Swift / ScreenCaptureKit

Demo video and download
The HapticMouse menu open in the menu bar

Mac app / Swift

HapticMouse

Gives any cheap USB mouse a haptic scroll wheel by borrowing the Taptic Engine in the MacBook's trackpad.

Swift / Released as a download

Website and download
The ZANYSURF browser agent working through a task

Browser agent / JavaScript

ZANYSURF Browser Agent

An AI agent that plans, clicks and fills in forms in the browser, fully local with Ollama or with other providers.

JavaScript / Ollama

On Microsoft Edge Add-ons
The Anbu Surveillance camera setup dialog

Computer vision / Python

Anbu Surveillance

Home security across several cameras: YOLO person detection, recorded clips and email alerts with a snapshot.

Python / YOLO / OpenCV

Source

03Experience

Building since 2024

Something shipped in most quarters. Newest first.

  1. Jun 2026 to now

    Eesa AI

    Phone app, add-ons, backend and web

    Working across the whole of an agent platform that businesses use every day.

    • Primary contributor to the iPhone and Android app in Flutter: chat, the Flow Board, and sharing a file to Eesa from any other app.
    • Primary contributor to five add-ons that plug into the agent as tools: Attendance, Drive, Invoices, Calls and QuickBooks.
    • Worked on workflows that run on a schedule and wait for a person's answer, role-based access for add-ons, and approval before the agent writes to an outside system.
    • 1,400+ commits across nine repositories in four months. The code is private, and I am happy to walk through any part of it.
  2. Oct 2026

    Tests, automated builds and downloadable releases across the open-source projects.

  3. Sep 2026

    SlideView, because no slide viewer did what I needed for revision.

  4. Jul 2026

    Aether, HapticMouse and Keyboard Strobe in one month. Restored the TVS Centra.

  5. Mar 2026

    ZANYSURF Browser Agent published on Microsoft Edge Add-ons.

  6. Feb 2026

    A school management system for a database course, and Quantum Classroom as a side project.

  7. Jul 2025

    Rebuilt a PC from scrap and made it a TrueNAS server.

  8. Mar 2025

    First shipped things: Birdfy, a browser game, and a YOLO home-security system.

04Hardware

I take hardware apart too

Two build logs, with photos of every stage. The same habit as the software: understand it from first principles, then build it.

A salvaged PC running open on a bench

Fig. 01Home server / 2025

Junkyard PC to home server

In first year I rebuilt a dead PC from e-waste: straightened bent socket pins, replaced a swollen capacitor and reflashed the BIOS. It now runs TrueNAS.

200 GBrecovered from a wiped disk
Read the build log The restored TVS Centra motorcycle at night

Fig. 02Motorcycle / 2026

TVS Centra restoration

My father rode this bike for 20 years. He and I brought it back with parts from a ₹3,000 scrapyard donor and a full AC to DC electrical conversion.

100,000 kmon the bike before we rebuilt it
Read the build log

05Education

Education

  • Third year

    SRM Institute of Science & Technology

    Undergraduate in Computer Science

    Chennai, India

  • 2024

    CS50 and Python on edX

    Where I started. The certificates are in one place.

06Contact

Have something that needs building?

I am open to internships and engineering roles, and glad to walk through anything on this page.

zanyanbuchelvan@gmail.com
LinkedIn GitHub