I built a Collaborative Data Canvas to end ' Multiple spreadsheet sharing issue'

Source: Dev.to

The Problem: The “Final_V12_updated.xlsx” Nightmare

We’ve all been there. You have a critical business decision to make, but the data is trapped in five different CSVs, three email threads, and a “static” dashboard that was last updated two weeks ago.

I spent months watching brilliant analysts get bogged down in VLOOKUPs and manual joins just to answer simple questions like, “Why is our LTV dropping?”

I decided to build a solution that treats data analysis like Figma: collaborative, instant, and visual.

Introducing AUM Data Labs

AUM Data Labs isn’t just another BI tool. It’s a Collaborative Canvas designed for speed.

The Technical Moat: Domain‑Aware Analysis

The core of the platform is a Python‑based Domain Engine. Instead of forcing users to write SQL or complex formulas, the engine uses proprietary logic to:

• Auto‑Detect Domains – It identifies if you’ve uploaded E‑Commerce, SaaS, or Biotech data.
• Instant Diagnosis – It runs benchmarked KPIs (GMV, Conversion Rate, etc.) and flags anomalies instantly.
• Secure Formula Evaluation – A custom evaluator ensures that while the logic is flexible, the environment remains secure against injection.

Key Features I’m Proud Of

• Visual Join Builder – Drag and drop columns to link messy datasets without a single line of code.
• Real‑time Collaboration – Built with a custom sync layer, you can see your teammate’s cursors moving across the canvas as you investigate together.
• Natural Language Querying – Ask “Show count of orders by category” and watch the chart render in real‑time.

The Architecture

• Frontend – React with a custom‑built canvas engine for high‑performance data manipulation.
• Backend – FastAPI (Python) orchestrating the heavy lifting of the analytical engines.
• State – Real‑time synchronization for multi‑user collaboration.

Why I Built This

I built AUM Data Labs during a period of intense professional transition. I wanted to prove that you can build enterprise‑grade technical excellence without a massive team.

My Goal: To make high‑level data diagnosis accessible to the people who actually make the decisions—Product Managers, Founders, and Ops Leads—not just the people who write the SQL.

I Need Your Brutal Feedback!

I’m launching the Free Tier (2‑Editor limit) today. I would love for this community to try and break it.

• Drop your messiest CSV.
• Try the Visual Join.
• Ask the NL bar a complex question.

Check it out here:

Demo: