Skip to content

LLM ready

/en/portfolio/operations-console.md

Title:
Data operations console

Data operations console

Internal Liquid AI product that became the source of truth for operations: the stage, owner, deadline and health of each client dataset sit in one place, for the team and for leadership. React 19 interface, Hono API, role-based access with Firebase and insights generated by Gemini.

Problem

Tracking each client's datasets was scattered and lived in Jira, with no view per dataset: the team and leadership could not see in one place which stage each dataset was in, who owned it and whether its deadline was at risk.

Decisions

I built the interface in React 19 and the API in Hono, with role-based access control in Firebase. Each stage has a deadline, and the console classifies each dataset as on time, at risk or late. Extraction and pre-normalization start from the interface itself, and Gemini generates insights about the processing queue.

Outcome

Delivered to production at Liquid AI as the source of truth for the client portfolio, with the history of each stage. It started the migration that moves tracking out of Jira. The code and the client names are not public.

---
title: Data operations console
url: https://nandomoreira.dev/en/portfolio/operations-console
collection: portfolio
tech:
  - React
  - Hono
  - Firebase
  - Gemini
  - TypeScript
date: 2026-04-01
---

# Data operations console

Internal Liquid AI product that became the source of truth for operations: the stage, owner, deadline and health of each client dataset sit in one place, for the team and for leadership. React 19 interface, Hono API, role-based access with Firebase and insights generated by Gemini.

## Problem

Tracking each client's datasets was scattered and lived in Jira, with no
view per dataset: the team and leadership could not see in one place which
stage each dataset was in, who owned it and whether its deadline was at
risk.

## Decisions

I built the interface in React 19 and the API in Hono, with role-based
access control in Firebase. Each stage has a deadline, and the console
classifies each dataset as on time, at risk or late. Extraction and
pre-normalization start from the interface itself, and Gemini generates
insights about the processing queue.

## Outcome

Delivered to production at Liquid AI as the source of truth for the client
portfolio, with the history of each stage. It started the migration that
moves tracking out of Jira. The code and the client names are not public.

Ready to share.