Hello, I'm Uba

AI Automation Builder

All systems

19 of 24AI systems / 19 of 24

Case study

Internal tool

Upwork Command Center

A personal live dashboard that streams new Upwork jobs every 30 seconds, auto-translates and summarizes them, and drafts human-sounding proposals on demand.

PythonUpwork API (OAuth)Kimi LLM APIHTML/CSS/JS

FIG. 001 / ARCHITECTURE

How the system connects

Live model

Problem

Finding and responding to good-fit Upwork jobs quickly matters for connects economics and response-time ranking, but manually refreshing the job feed, reading each listing, and writing a proposal from scratch for every promising one doesn't scale.

System

A Python watcher polls the Upwork API (OAuth) every 30 seconds, deduplicates new jobs into a local feed, and pre-generates a Turkish summary/translation for each via an LLM. A lightweight Python server exposes a tabbed HTML dashboard (live feed + overview) with endpoints to view full job details, generate a draft proposal, submit an approved application, and track per-job and aggregate LLM token cost.

FIG. 002 / HOW IT WORKS

How it works

A single-user automation built to compress the Upwork job-hunting loop: instead of refreshing a feed and re-reading listings, new jobs arrive pre-summarized and pre-translated, with a one-click path from "interesting" to a drafted, human-toned proposal ready for a final human approval before it's sent. It also tracks its own LLM token spend so the automation's running cost is visible, not a surprise.

Stack

What it runs on

PythonUpwork API (OAuth)Kimi LLM APIHTML/CSS/JS

Want a system like this for your team?

We start by mapping one workflow on a 30-minute call.

Message me onWhatsAppTelegram