DEV Community

Achyut Srivastava
Achyut Srivastava

Posted on Originally published at luxurai.in

Meet LuxurAI: The Autonomous AI Operating System Built by a 14-Year-Old Founder in India

Meet LuxurAI: An Autonomous AI Operating System Built in India

Over the past year, we have been building LuxurAI — a unified AI platform and autonomous operating system designed to make AI agentic tools super fast, reliable, and accessible without expensive foreign dollar subscriptions.

Here is an authentic breakdown of who is building LuxurAI, how our engineering stack works under the hood, and what we are shipping in our public beta.


👥 The Core Team & Founding Story

LuxurAI is independently engineered and bootstrapped with ** VC funding** out of India:

  • 🚀 Achyut Srivastava — Founder & Lead Architect (Age 14) Designs the core multi-model orchestration, AST code-healing loops, and the in-house Neo v0.1 vision perception model.
  • 🤝 Shubham Dangi — Co-Founder Collaborates on platform infrastructure, system scaling, and community distribution.

We started LuxurAI because existing AI developer tools are locked behind heavy –/month subscriptions. We wanted to build something radically accessible: ₹0.25 per Luxur Credit (1 LC = ₹0.25), backed by real agentic autonomy.


🛠️ Deep Architectural Breakdown

1. In-House Neo v0.1 Vision Engine (LuxurAI Cowork)

Our desktop coworker runs an in-house vision model (Neo v0.1) on a dedicated dual-GPU cluster. It captures screen frames at 15 fps and performs localized OCR and UI element bounding box detection.

`python

Conceptual perception loop for Neo v0.1

async def perceive_and_act(frame_stream):
async for frame in frame_stream:
ui_nodes = await neo_vision_model.detect_bounding_boxes(frame)
user_intent = classify_intent(frame, ui_nodes)

    if user_intent.requires_destructive_action:
        # Human-in-the-loop safety gate
        await request_user_confirmation(user_intent)
    else:
        execute_win32_action(user_intent.action_plan)
Enter fullscreen mode Exit fullscreen mode

`

2. Multi-Model AST Parsing & Merge Validator (LuxurAI Coder)

LuxurAI Coder does not just blindly emit code; it builds an Abstract Syntax Tree (AST) map of the repository, creates targeted diff blocks, and runs verification checks (bracket balance, pyflakes, and test suites) before proposing modifications.

Component Technology Primary Function
Vision Perception Neo v0.1 (Dual GPUs) Real-time screen UI detection & OCR
Code Agent Multi-Model AST Engine Atomic patch generation & self-healing
Safety Dial Win32 / Overlay UI Human-in-the-loop approval gate
Backend Core FastAPI + SQLite/Postgres High-concurrency event routing & WebSockets

🌐 Transparent Public Beta & Roadmap

LuxurAI is currently in active public beta:

  1. Live Cowork Perception: Continuous improvements for multi-monitor setups.
  2. Coder AST Optimization: Pruning token bloat on large monolithic repos.
  3. Open Beta Testing: Try it free at luxurai.in.

We welcome feedback, architectural critiques, and edge-case reports from fellow developers!


Built with ❤️ in India by Founder Achyut Srivastava and Co-Founder Shubham Dangi.

Website: https://luxurai.in

Top comments (0)