|

• Monday: drove my motorcycle up to Gardiner to meet up with H, see his new house, and chat about some potential work, received blue bottle shipment
• Tuesday: worked in my new watercolor sketchbook, got a few things done for the AP
• Wednesday: Made some more progress getting up to speed with the AP’s elections systems, had a meeting with S from val.town about potential collaboration, worked on some bugs with connectology
• Thursday: Finished 2.5 PRs for the AP, worked on getting scrapbook-core running in the cloud, created script to update Cloudinary photos with computer vision descriptions
• Friday: went to Warwick to work from Ben’s studio, did an AP election test call, recorded some VOs for potential videos
• Saturday: worked on cutting together a youtube video on scrapworld, worked on rewriting a markdown parsing system for my blog

💻 Code & Creations

Connectology

Slowly and steadily pushing Connectology to a place where I can demo it and share with other people, and most importantly be proud of it.

Worked a bit on fixing the login flow, tweaking and simplifying styles across the board

DBPedia Integration

Cloudinary Annotation

Created a basic repo / script that requests the last week of Cloudinary photos and sends them to gpt-4o to generate descriptions. Went with this vision prompt:

Describe this image in detail, breaking it down into the following aspects: 1) Main subject matter, 2) Visual elements like composition and setting, 3) Any visible technical content (such as code or devices) with a focus on the general purpose or functionality rather than specifics, and 4) Inferred context, such as time of day, season, or mood. Finally, generate 5-20 relevant tags based on the description, aiming to include tags like 'code,' 'photography,' 'art,' 'nature,' '3d,' or other relevant terms from a wide range of topics. Try to keep your description under 400 words. Avoid lists. At the end, make some inferences/guesses as to what the user is doing, if applicable. Keep your answer as information-dense and succinct as possible. Try to capture the overall gist of the image as well as the specifics. Never respond in the negative.

Scrapbook-core

Working on getting this deployed to fly.io on a cronjob so I don't have to run it by hand, which means my bookmarks are almost always out of sync lately.

Updated blog

I worked on creating a custom markdown->json parser I could use to power my blog, since the bug that's preventing me from running my blog with @nuxt/content hasn't been fixed yet and I am extremely impatient.