๐ Today’s Picks
Faker: Generate Realistic Test Data with One Command
Problem
Creating realistic test data manually is time-consuming.
Solution
Faker generates authentic-looking test data with single-line commands.
Key features:
- Realistic names, emails, and addresses
- 50+ language locales (en_US, vi_VN, etc.)
- One-line profile generation with custom fields
Persist Agent State Across Restarts with LangGraph Checkpointing
Problem
Checkpointing is a persistence layer that maintains agent workflow state between executions.
Without checkpointing, agents lose all state when systems restart, requiring users to start over with new conversations.
Solution
With LangGraph’s checkpointing, you can persist agent state to databases, enabling:
- Conversation continuity through restarts
- Same conversation accessible from any application instance
- Flexible persistence with PostgreSQL, SQLite, or MongoDB backends
โ๏ธ Weekly Finds
git-who [Data Engineer] – Git blame for file trees – visualize code authorship and contributions across entire directory structures
nanochat [LLM] – The best ChatGPT that $100 can buy – minimal, hackable LLM implementation with full training pipeline
ManimML [ML] – Animate and visualize machine learning concepts with Manim – create neural network visualizations and educational content
Looking for a specific tool? Explore 70+ Python tools โ
Stay Current with CodeCut
Actionable Python tips, curated for busy data pros. Skim in under 2 minutes, three times a week.


