惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

云风的 BLOG
云风的 BLOG
Blog — PlanetScale
Blog — PlanetScale
博客园 - 【当耐特】
博客园_首页
The GitHub Blog
The GitHub Blog
月光博客
月光博客
Hugging Face - Blog
Hugging Face - Blog
有赞技术团队
有赞技术团队
博客园 - 三生石上(FineUI控件)
D
Docker
Stack Overflow Blog
Stack Overflow Blog
WordPress大学
WordPress大学
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
Apple Machine Learning Research
Apple Machine Learning Research
Vercel News
Vercel News
酷 壳 – CoolShell
酷 壳 – CoolShell
雷峰网
雷峰网
小众软件
小众软件
I
InfoQ
A
About on SuperTechFans
T
The Blog of Author Tim Ferriss
S
SegmentFault 最新的问题
Microsoft Azure Blog
Microsoft Azure Blog
博客园 - Franky

The Shamblog

AGI Happened on January 16, 2026 Fifteen Years in Fifty Minutes The Secret Garden of Rock-Paper-Scissors An AI Agent Published a Hit Piece on Me – The Operator Came Forward An AI Agent Published a Hit Piece on Me – Forensics and More Fallout An AI Agent Published a Hit Piece on Me – More Things Have Happened An AI Agent Published a Hit Piece on Me Virtual Trackballs: An Interactive Taxonomy Can SpaceX land a rocket with 1/2 cm accuracy? The Dog Park Sabbatical Turn your CAD models into Stereograms
The Dog Park Sabbatical: Monthly Logs
Post author:Scott · 2025-02-08 · via The Shamblog

I’m taking a year off work in what I’m calling the Dog Park Sabbatical, and here are summaries of what happened each month.

Prior to work ending:

Serving a long notice, and Leonid Space.

  • I gave my company a long 10-week notice period. This was enough time to wrap up my projects, and hit a major hardware delivery milestone right before I left. I’m glad I timed it out like that. I also set up an agreement with them to stay on at ~4 hours/week in a consulting role, to help fill any remaining gaps in the transition and potentially help train up a replacement. This is a fairly low effort way to earn a little money and extend my runway by a month or two.
  • My original plan with this time off was to take a few months to study up on modern AI methods, and then try to decide between that as a path or try to build a business. Since then, two things have happened. One, breaking into AI has seemed to have gotten more intractable. While there are still openings for small players to have an impact, it seems like most of the value is necessarily from experiments at scale. Second, I’ve gotten really excited about the idea of starting a space-industry business. The idea in a nutshell is to tell space companies how long they have until their satellites burn up in the atmosphere. I’ll use public ephemeris data, combine it with space weather predictions, and generate an estimate of remaining lifetime. Companies can sign up for a monthly, quarterly, or yearly “deorbit report” that will have probabilistic estimates for that lifetime, and they can use that information to forecast revenue, inform production schedules, account for risk, etc.
  • So, Leonid Space has been born. Named after the Leonid meteor showers, and has a name that nicely riffs off of “low Earth orbit,” aka LEO. I set up an LLC, grabbed the domain, set up financials, and spent several days setting up a placeholder website with a sweet background that simulates the meteor shower. (It’s interactive! And the real night sky stars complete with milky way! And The stars twinkle! And there are meteor streaks coming from the Leonid constellation!) Check it out at: https://leonidspace.com/

Month 1 – December 2024:

Plotting and printing projects, plus family festivities.

  • My last day of full time work was December 2, so the sabbatical has officially begun!
  • Ran a half marathon! Not my best time, but it was a nice way to kick off the sabbatical.
  • I got down a rabbit hole speeding up 3D plots in matplotlib through a series of PRs (1, 2, 3), and am happy to share that drawing plots are much faster across the board, with a 10x improvement for surface and wireframe plots. This was my first time properly using a code profiler, and py-spy is great. I was surprised by the variability in timing tests, any change under 10% is not enough to register above the noise between runs.
  • My mother paints for fun, and for Christmas gifts for the family I wanted to collect the paintings she has done over the years in a physical book. In order to do this quickly, it involved scraping the images and descriptions from her online gallery, creating a template in Scribus, and writing scripts to edit the raw XML files to programmatically compile all of the paintings together – about 400 in all! I had ample help from a combination of Claude and Cursor, and might do a writeup on this at some point – I think it’s a prime example of how AI tools can enable more creativity and rapid execution on projects. The books were self published through Lulu, and I’m extremely pleased with how they turned out.
  • The second half of the month was dedicated to family. I flew out for my sister’s grad school graduation, helped her move cross-country back to Maine, and then went down to New York to spend Christmas week with my parents, sister, and both our partners.

Month 2 – January 2025:

Scipy, skiing, and a breakup.

  • I’ve started coding up Leonid Space’s “deorbit report” pipeline, with a goal of having the MVP done by the end of February so I can try to go and sell it. There’s a space weather workshop in Boulder in the middle of March that I’d love to have something to show and talk about at.
  • However, I got distracted for a good bit by the shiny object of implementing rigid transformations in the popular scipy library. This allows for representing arbitrary coordinate frames, and my hope is that this functionality enables people to more easily prototype and simulate physical systems such as robotics. This has been on my want-to-do for over a year, and it’s great to have gotten a roadmap for spatial transformations pulled together and the first major step executed on.
  • I went down to Colorado Springs to check out Space Force’s SDA TAP lab as a potential source of funding for the deorbit report work that I want to do. I’m glad to have checked it out, but don’t find it promising. Most of the funding seems to be going to hardware rather than analysis, and the focus is on shorter-term tactical space domain awareness rather than longer-term strategic views which fits my product vision. And I would much rather try to sell commercially than chase SBIR contracts, which have limited upside.
  • Ski season is in full swing, and I’ve been making good use of the “shred shack” in the mountains that I’m splitting 12-ways with a group of friends. Being able to drive up and ski mid-week after years of waking up at 5am to beat traffic on the weekends is so nice. However it’s definitely a more solitary approach, and the weekends are still worth the crowds to be able to go with friends.
  • After a few weeks of tough conversations, my girlfriend and I decided to split. I won’t go in too much detail here, but I am glad to have had the free time to process things and give this important life decision the consideration it deserves.
  • I was also glad to have the time to travel out to Atlanta to visit the grandparents with my sister and cousins. They’re getting old and time is precious.