AI-first product studio — New Delhi, working worldwide
Ship the AI productyou've been scoping for months.
Machine learning, AI agents, voice and video, websites and mobile apps — from one senior team that has been training and deploying models since 2019, not since ChatGPT.
Free 15-minute teardown · reply within one business day
AI Voice Studio — our own speech platform, built and operated by us in 23 languages.
Try it live — ai.rekvon.in →
You would be working with me directly — I read and answer every enquiry myself.
Aditya Pratap Singh, New Delhi, India
01 — Selected work
Two things we built and run ourselves. Client case studies land here as they ship.
AI Voice Studio — live product →
AI Voice Studio — mobile →
Rekvon.in — mobile →
Rekvon.in — this site →Want to be the third one here?
We write up every project we ship. Yours could be the next case study on this page.
02 — Services
Not sure which of these you need?
That is what the free 15 minutes is for. Describe the problem and we will tell you the shortest route to it.
One studio, every surface
Agents, voice, video, web, and mobile from a single senior team — one accountable partner instead of a vendor relay race.
Fast first version
A working version in weeks, not quarters. Momentum from week one, a demo every week after.
Conversation, voice, or flow
Chat agents, voice agents, or silent automations — we build whichever actually fits the job, not whichever is trendy.
Scales as you grow
One build becomes a platform on the same foundations — no rebuild when it's time to grow.
03 — One studio, your whole stack
We plug into everything you already run on.
AI & models
OpenAI, Claude, Gemini, or open-weight Llama — plus custom voice cloning and AI video pipelines built on top.
Apps, payments & backend
Stripe and PayPal checkout, Firebase and Supabase backends, shipped to the App Store and Play Store.
Web & workflow
Modern web on Vercel, commerce on Shopify, and the everyday tools your team already runs on.
04 — Our product
AI Voice Studio.
We don't just build for clients — we ship our own. AI Voice Studio is Rekvon's in-house speech platform: voice cloning, narration in 23 languages, live in production today.
Screenshot of the live product — click to open
05 — The numbers
Trust the numbers, not the adjectives.
AI voice generated in production
140+ hrs
Languages in our live voice engine
23
Avg. idea → first live version
5.8 wks
94%
< 24 h
Enquiries answered within one business day
06 — Track record
We were training models before prompting was a job.
Most studios selling AI today started when ChatGPT did. Ours started in 2019 with machine learning and deep learning — classifiers, NLP, Kaggle competitions, a model deployed behind a cloud API — and it has all been public the whole time. Open any of it.
- 2019
- Building machine learning since
- 81
- Public projects you can read
- 6
- Languages shipped in production
Classifiers, NLP and a deployed model API — before generative AI existed
Not forks — things built from scratch
Python · TypeScript · JavaScript · Go · Dart · Shell
Machine learning
Classification, regression and forecasting on your own data — trained, evaluated, and deployed behind an API rather than left in a notebook.
Deep learning
Neural networks for language, vision and audio where classical methods run out — including the training and evaluation loop, not just inference.
NLP & language systems
Text classification, similarity, extraction and retrieval — the groundwork underneath every retrieval-augmented product that actually works.
Generative AI & agents
LLM agents, voice and video systems, and model integration — built on the evaluation discipline the older work taught us.
ViralLabs
Source
Full-stack AI platform for YouTube creators — script generation, thumbnails with face cloning, and voiceovers on GPT-4 and Gemini. React 19 + FastAPI.
Open ↗smart-postgres-mcp
Source
An MCP server that lets an AI assistant query and reason over a Postgres database safely — the same integration pattern we build into client products.
Open ↗Fake News Classification
ML · 2020
A machine-learning classifier that flags likely-fake articles — one of several NLP and Kaggle builds from 2020, three years before generative AI became the default answer to everything.
Open ↗IriisApi
ML · 2020
An ML model deployed to a cloud API in 2020. Training a model is the easy half; serving one is where most machine-learning work quietly dies.
Open ↗Study Buddy
Live
A deployed TypeScript study app — AI-assisted learning, shipped and running.
Open ↗The Quran Guide
Live
A deployed reading and search interface built in TypeScript, live for anyone to use.
Open ↗
07 — How we work
Discover
We map the problem, the constraints, and what 'working' actually means for you.
Design
Interface, architecture, and model choices — decided together, in the open.
Build
Short cycles, working software every week, no black box.
Ship & run
Deployment, monitoring, and iteration after launch. We stay accountable.
Week one starts on a Monday.
Tell us what you are building and we will tell you which of these four weeks yours actually needs.
08 — What it costs
Real numbers, before you get on a call.
Most studios make you sit through a discovery call to find out whether you can afford them. Here is the range up front, so you can decide in thirty seconds instead of thirty minutes.
Sprint
$2,500from, fixed scope
One clearly-defined thing, shipped properly.
2–3 weeks
- A conversion-focused site, an agent prototype, or an AI integration audit
- Direct access to the engineer building it — no account manager
- Working software demoed every week
- Source code and infrastructure handed over, yours to keep
Build
Most common$8,500from, fixed scope
A real product, in production, with users on it.
6–8 weeks
- Everything in Sprint
- Full product build — web, mobile, or an agent that runs unattended
- Machine learning or LLM work where the problem actually needs it
- Deployment, monitoring and a fortnight of post-launch fixes included
- Payment gateway, auth and dashboards where the product needs them
Partner
$4,000per month, ongoing
You need a product team, not a project.
Rolling, cancel with 30 days
- Everything in Build, continuously
- A standing share of our capacity each month
- Roadmap and technical direction, not just execution
- We operate what we ship — uptime, cost and latency are our problem
- Priority: your work goes to the front of the queue
Fixed scope means fixed price — if we underestimate, that is our problem, not your invoice. Every number above is a starting point for work of that shape; you get a real figure on the first call, before you have spent anything.
Free 15-minute teardown · reply within one business day
09 — Who we work with
For founders and teams who ship.
You get agency-grade quality without agency overhead, and more accountability than any marketplace freelancer — a senior team, founder on every project. Even this site is our work: read how it was built.
Who
Founders and small-to-mid teams who need AI shipped into a real product — not a proof of concept. If you have a workflow worth automating or a product worth building, we're a fit.
How
Fixed-scope builds for defined projects, or a monthly engagement for ongoing product work. Short cycles, working software every week, direct access to the people building.
Budget
Projects start from $2,500. No generic rate card — your project gets a real number on the free first call, and every inquiry gets a reply within one business day.
10 — Where we work
Working worldwide, shipping from Delhi.
Remote-first from New Delhi, covering US, European, Middle East and APAC hours. Senior quality with real timezone overlap — without Bay Area pricing.
11 — Straight answers
The questions you'd ask a friend about us.
Including the ones that are awkward for us. If something here worries you, that is worth knowing before we talk, not after.
Who actually does the work?
Aditya Pratap Singh — the founder — writes the code, and brings in senior specialists when a project needs a discipline he does not cover. You will not be handed to a junior team after signing, because there is no junior team. That is the trade: you get less headcount and more of the person you actually evaluated.
You have no client logos. Why should I trust you?
Fair. We have not put fake ones up either. What we can offer instead is evidence you can check yourself: a live AI product at ai.rekvon.in that we built and operate, this website, and seven years of public code going back to 2019 — including the machine learning work from long before AI was fashionable. Read any of it before you spend a rupee.
What happens if you disappear halfway through?
You hold the keys the entire time. Code lands in your repository, infrastructure runs in your accounts wherever possible, and you get the credentials as we go rather than at handover. If we vanished tomorrow, another competent engineer could pick it up on Monday — we write for that person.
Why are Indian rates cheaper? What is the catch?
Cost of living, not cost of quality. The same work billed from San Francisco carries Bay Area salaries and Bay Area office overheads; ours does not. The catch you should actually check for is seniority — plenty of cheap work is cheap because a junior is learning on your budget. Ask us who is writing your code, and check their public work.
Do I own the code?
Yes, once it is paid for. That is written into every engagement. We retain our own pre-existing tools and libraries, and we tell you clearly and in advance if any of those are involved so you are never surprised by a dependency you cannot control.
Do you only do generative AI?
No — and that distinction matters. We do machine learning, deep learning and NLP as well as LLM agents: forecasting, classification, computer vision, recommendations. Sometimes the honest answer to a problem is a small trained model that costs cents to run, not a large language model that costs pounds. We will tell you which one you need.
How do we work together across timezones?
We are in New Delhi and overlap with US mornings, all of Europe, the Gulf and APAC. In practice you get a written update whenever something meaningful ships, a working demo every week, and a reply within one business day — not a standing 6am call neither of us wants.
What if the project turns out to be smaller than I thought?
We tell you on the first call, before you have paid anything. Talking someone out of a build they do not need has cost us work and earned us referrals. If the whole thing is a weekend of scripting, we would rather say so than bill you for a quarter.
How do we start?
Send a few honest sentences through the contact form. You get a reply within one business day, then a free fifteen minutes that ends with a real scope and a real number — whether or not you hire us. No deck, no discovery invoice, no sales sequence afterwards.