First weekend with the @CFB_Data starter pack and have been using @claude_code and @databricks to "chat" with my raw data and make dimensional tables. First I thought MCP would get rid of data modeling but now am realizing that modeled data makes it better
Forget Ralphing your code , I am Globex Homering my coding agents (standing there useless, asking if they could work a little faster every few minutes)
Focus Friend, but instead of knitting virtual socks, your Friend makes a small donation to your favorite football teams NIL after you watch a short ad. This will boost work productivity in the SEC and B1G 10 states by 1000%. How can we make this happen @Blake_Lawrence ? :)
#claudecode lesson of day, from Claude Code: Always leads with constraints research, not solution building.
Just another example of how the future is Context Engineering, not Prompt Engineering
@mihai673@sidbid I had same thought but the initial agent sometimes rushes and makes assumptions on code even after sharing documentation. I had agent #2 who was Sr Data Engineer review code and reject changes because it was more thorough. It is weird concept but it’s all the same Claude instance
@BosonJoe@thepatwalls I started the same path you did, vs code and sonnet was ok but a lot of back and forth with the IDE. First day using Claude Code in vs code terminal I made a boilerplate website in 30 minutes and 2 POCs by end of day. Switch to CC, you won’t regret it.
Also learned how to use the markitdown package to convert the PDF instructions included in the starter pack into something that Claude Code could read and help inform the modeling process. They have a MCP so could use that in the future as well. https://t.co/70qU0hwFFk
First weekend with the @CFB_Data starter pack and have been using @claude_code and @databricks to "chat" with my raw data and make dimensional tables. First I thought MCP would get rid of data modeling but now am realizing that modeled data makes it better