Well hey there - you’ve probably heard the news, but I’ve left Clearbit. It was a good 5 year stint where I learned a lot, built some cool things, and did lots and lots with data.
5 years is a long time in tech, and far longer than I was expecting to spend at Clearbit in the first place. My original plan was to use it as a launch pad, to get out to SF, to level up my skills, and to launch into the next thing - 3 years tops. A combination of the people I worked with, problems I worked on, and the pandemic kept me there past my expected horizons.
So, lets get back to the plan and launch into something else.
Theres a ton of ideas knocking around in my head and this is where I’m going to document what I’m picking at and where its going. Think of this like the “build in public” ethos, but more targeted and less screaming into the void that is twitter et al.
So what is next?
AI is the topic thats on everyone’s mind right now and for good reason. A large number of tasks that required humans just got automated - exactly which ones AI is able to mechanize is still an open question and one that I’m looking at exploring.
The current project - I have a few years worth of podcast data, damn near all of it to be honest, or at least all the shows that people care about. After about a weeks worth of hacking I’ve set up a system that can ingest, translate, transcode, speaker detect, and content extract about 500 hours of audio content at day1. Right now I’m just compiling this data, but I’m starting to get to a point where I can ask a few questions of it:
What are people talking about?
What ads are running on what content?
What are non-english language podcasts talking about?
Is there trends to any of the above?
The actual product that this may be is still fuzzy to me, but I’m operating on the theory that creating a dataset somewhere that no one else inherently has value - especially to the non-technically inclined.
Have something you’d like to know about this dataset? Just ask, we’re all friends here.
The lights in my apartment flicker when this pipeline runs at full bore