← All work
Case study

Axiora

A generative dashboard engine that infers KPIs directly from raw CSV and Excel files with no manual setup, and answers questions about the data in plain language.

Visit the live site →
Role
Software Engineer
Company
Axiora
When
Jul 2025 — Mar 2026
Stack
Python · LangChain · FastAPI · Flutter
12pilot companies off manual analysis
2products, desktop and web
Zeromanual setup per dataset
The engine · raw file to dashboard
From file to answer
  1. Raw CSV / ExcelNo setup required
  2. KPI inferenceThe engine decides what matters
  3. Generated dashboardBuilt from the data
  4. Free-text re-scopingChange the dashboard in plain language
  5. Chat over the dataAd-hoc questions

The problem

Axiora’s job was to replace manual data analysis: drop in a CSV or Excel file, get a dashboard.

What I built

KPIs inferred from the raw file

The engine reads raw CSV and Excel files and works out which KPIs matter, with zero manual setup.

Natural language across the product

Users re-scope a dashboard by typing what they want, and can chat with the dataset for ad-hoc questions.

Two products on one backend design

I owned the backend architecture of both products, the desktop app and the website, end to end.

The calls

Call 1 · Ownership

One engineer, whole backend.

Owning the architecture of both products end to end meant one set of decisions, made once, with no hand-offs between them.

Call 2 · Interface

Plain language instead of configuration.

Inferring KPIs and taking free-text instructions means there is no configuration step to learn.

Outcome

Axiora replaced manual data analysis for 12 pilot companies.