Yu Dong

Data Science & Analytics, Data Storyteller

2 minute read

Reading List in Past Two Months

Here are the great articles I came across in the past two months.

Data Science & Analytics

  1. Meta: A Blueprint for Valuing Content When A/B Tests Are Not an Option: How the Meta team uses DoubleML to measure the incremental impact of interacting with certain content.
  2. The Death of Dashboards: The structural problem today is that dashboards answer the questions analysts anticipated, not always the questions users actually need to ask at the moment. And how LLM has transformed the workflow.
  3. Storytelling is the wrong skill for Dashboards — what to learn instead?: why cannot we have storytelling in a dashboard, and what is that often overlooked critical skill specifically benefiting dashboard creators?
  4. How to Use a Time Series Foundation Model for Anomaly Detection: Pretrain a large time-series foundation model (TSFM) and apply it across many time series, out of the box for forecasting and then anomaly detection. Use Chronos-2 from AWS as an example.
  5. Data Bridge: How Netflix simplifies data movement: Data Bridge is Netflix’s unified control plane that simplifies and standardizes data movement across our diverse ecosystem of data stores.

Data Career

  1. I’m a Data Analyst Who Barely Does Analytics Anymore.: The traditional role that comprised writing SQL queries and building dashboards might no longer be in high demand. But hybrid roles will emerge.
  2. What Does Analytics Leadership Look Like in the Age of AI?: Transform from manaaging insights to managing intelligence.
  3. 10 Data + AI Predictions for 2026: Great observations of data industry trends related to AI.

AI and LLM

  1. A Complete Guide to Claude Code: Hooks: The Hook mechanism lets you register processing logic that must run during Claude Code’s execution
  2. Claude.md vs Skills vs Subagents vs Hooks: How to Choose the Right Claude Code Customization Layer: CLAUDE.md tells Claude how to behave all the time, Skills tell Claude how to think in specific situations, Subagents let Claude do work in a separate context, Hooks run automatically when events happen, and MCP gives Claude access to outside systems.
  3. Your AI Has Amnesia. Claude Subagents Are the Cure.: The model’s attention degrades as the history grows, and how to use subagents to solve it. Detailed walkthrough of what is subagent, how to invoke them, subagent file structure, and common use patterns.
  4. CLAUDE.md Best Practices: 10 sections to Include in your CLAUDE.md.
  5. How I Continually Improve My Claude Code: Tips to improve Claude Code efficiency, including daily cron job of a Claude Code conversation review skill, improving human interaction with coding agents, letting the agent ask you questions, etc.
  6. Self-Service Analytics, According to Anthropic: Analytics accuracy is a context and verification problem, not a code generation issue. How data foundations, sources of truth, skills, and validation help.
  7. 5 Ways AI Is Reshaping Data Analytics (And what you can do about it): The role is shifting from data builders to data architects and storytellers.
  8. Why Traditional ETL is Dead: The Rise of Agentic Data Engineering: How to build AI-ready data foundations without friction to replace traditional ETL.
  9. How Image Generation Actually Works: A high-level walkthrough of the two image generate methodology families: sculpting from noise, and writing an image like a sentence.
  10. Inside Meta’s Home Grown AI Analytics Agent: how to build a AI agent that performs routine data analysis tasks autonomously.

You May Also Enjoy