This post was written using DeepSeek-R1 AI. The tool seems to be surprisingly precise for writing!

This post highlights three weekend projects—small tools built to solve real problems. No boilerplate, no grand missions. Just focused code for niche workflows: managing tasks, simplifying AI integrations, and analyzing market data. Built with modern tools (FastAPI, Python, lightweight JS) but designed to stay simple. Think of these as blueprints, not polished products. Let’s dive in.

1. NanoSaaS: A Microservice Platform for Task Control:

What it does: A lightweight system to manage computational tasks with a credits system.

Key features:

- Credits system: Users get a number of credits to run tasks, preventing resource abuse, and allowing you to sell your software as a service.

- Real-time monitoring: Track task progress live without refreshing the page.

- Google SSO: Secure login without managing passwords.

- Stack: FastAPI (backend), Celery (task queue), Alpine.js + Tailwind CSS (frontend).

Why it matters: Perfect for small teams or solo developers who need to track jobs without overcomplicating things. No bloat—just tasks and quotas.

→ GitHub repository: https://github.com/carlosplanchon/nanosaas/

2. FastAPI AI Assistant: A Chatbot That Leverage OpenAI Assistants API:

What it does: A minimalist FastAPI app that wraps OpenAI’s Code Interpreter into a real-time chatbot.

Key features:

- Streaming responses: Answers appear word-by-word via Server-Sent Events (SSE), no waiting for full completion.

- Simple integration: FastAPI acts as a clean middleware layer between OpenAI and the frontend.

- Stack: FastAPI (backend), Alpine.js (frontend), Tailwind CSS (styling).

Why it matters: This is a small template for adding conversational AI to apps.

→ GitHub repository: https://github.com/carlosplanchon/fastapi_ai_assistant

3. MeasureVolume: Gauging Market Taker Activity:

What it does: Analyzes order book snapshots (like those from crypto exchanges) to estimate trading volume and liquidity changes.

Key features:

- Order book diffing: Compares consecutive snapshots to infer executed trades.

- Sample datasets: Includes toy data to experiment with.

- Scripts: A script to gauge market takers activity.

Why it matters: For traders or researchers, this offers a lightweight way to study market dynamics.

→ GitHub repository: https://github.com/carlosplanchon/measurevolume

Sometimes, all it takes is a weekend and a focused idea to bring something new to life! 🛠️