RoleEngineer by training, AI nerd by choice
Based inTN, India
Local time IST
StatusOpen to projects

Boopathi Thangamani

I'm an AI Solutions Architect in TN, India. I design and integrate AI into any business: AI that runs every day and tells you when it's working. I also build my own products, all in beta right now, so the advice I give clients comes from work I do every day.

§ 01 Work · 2026

Four products, all in beta.

Each one shown through the screen that matters most. The screens play as you reach them; press Replay to watch again.

01 / 04 Azhagi MetaHuman · vision · voice 2026 · Beta, in development

An AI influencer you can call, face to face.

Azhagi is a live MetaHuman on a video call. She sees you through the camera, hears you, talks back in real time and performs any move you ask for. Every reply and every gesture happens live, on the call.

She is built as an AI influencer: a presence an audience can call and talk to, not just watch. She is in beta and not yet released.

  • 0Live senses: sees, hears, speaks, moves
  • 0Reply latency
  • BetaIn development, not yet released
Azhagi / video call · live
02 / 04 Boona Cue Text → animation · any engine 2026 · Beta, in development
Boona Cue / editor · desktop

Words in. The exact performance out.

Boona Cue turns words from any AI into animation for AI characters. Tell it "blink twice, then nod three times" and the character blinks exactly twice and nods exactly three times, on any engine.

The language model decides what to say. Cue makes sure the body keeps its word: the exact performance, every time, nothing extra. It is in beta and not yet released.

  • 0Blinks asked for, blinks delivered
  • 0Nods asked for, nods delivered
  • 0Cues dropped or invented
03 / 04 Thali React Native · iOS · TestFlight 2026 · Beta, in development
Thali / this week · iPhone

Your meal rules, turned into a week of South Indian food.

Thali is a React Native app. You set the rules of your meal planner, and it turns them into a weekly South Indian meal plan, sized for the number of people eating.

It is in beta on iPhone through TestFlight as in.boona.thali, not yet on the App Store.

  • 0Days planned at once
  • 0Slider sizes every portion per person
  • iOSBeta on iPhone via TestFlight
04 / 04 Boona Fit Stack · React Native · Android · iOS 2026 · Beta, in development

A fitness plan that fits the person, not the template.

Boona Fit is a personalised fitness app. It plans the programme, guides the exercising, and tracks what actually happened. It is in beta and not yet released.

AI-powered personalised fitness, meal and supplement planner.

  • 0Modes: plan, exercise, track
  • 0Plan per person, not per template
  • BetaIn development, not yet released
Boona Fit / this week · app
§ 02 Approach

How a project runs.

Listen first, prototype on real data, instrument before shipping, hand over properly. Every date comes as a range.

Working AI, measured from day one, and a team that can run it without me.

  1. 01

    Listen first

    What does the business actually do all day? I start there, not with the AI wish list.

  2. 02

    Map the hours

    Where the time goes, written down. We pick the one flow worth automating first and the number we expect it to move.

  3. 03

    Prototype on real data

    A working slice within days, on your data, not a slide. If it does not earn its keep, we stop early and cheap.

  4. 04

    Instrument before shipping

    Telemetry first. I once ran production without it and debugged the outage by feel. Never again: if we cannot see it, we do not ship it.

    Rule
  5. 05

    Dates come as a range

    Targets I once set to a single day got missed. Now every date is a range, and the range holds.

    Rule
  6. 06

    Hand over, properly

    Documentation, dashboards, and a team that can run it without me. I am the architect, not the dependency.

The shape of a project

Start to hand over · length depends on the job
Understand and prototype
First quarter
Build
Middle half
Instrument, alongside the build
Never after
Launch and hand over
Last quarter
Dashed tail = the range. Every date comes as a range, and the range holds.Every project has this shape; only the length changes
§ 03 Experience

Ten years, from web apps to AI.

Two rules I learned the hard way: monitor before launch, and give every date as a range.

10years
  • A bank's customer onboarding5 days → about 1 day
  • An online store's load timeabout 10 s → 3 s
  • A new store live on a food-ordering platform48 h → 24 h
  • Time to put a release liveabout 50 % less
  1. 2026 — nowCurrent

    Independent, as Boona

    Boona · TN, India

    AI automation for clients (anonymised). Building Azhagi, Boona Cue, Thali and Boona Fit. Publishing on LinkedIn. Boona AI Agency is forming behind it.

    AzhagiBoona CueThaliBoona Fit
  2. 2022 — 2025

    Lead Software Engineer at a product company

    Ran production without telemetry. Debugged the outage.

    Rule since · Telemetry first
  3. 2018 — 2022

    Senior Software Engineer on a large e-commerce application

    Set targets and dates that were missed.

    Rule since · Realistic dates, with a range
  4. 2015 — 2018

    Trainee Software Developer, then Software Engineer

    Web apps and React front ends. Includes the 2018 food-ordering platform: store onboarding brought from 48 hours to 24, with a product team of eight.

    Web appsReact
§ 04 Contact · TN, India

Tell me what the business does.
I'll tell you where AI fits.

One email is enough to start. No deck needed.

Email
Studio
TN, India
Local time
IST
Working as
Independent · Boona AI Agency forming
Opens your email app with the brief filled in.