About this archive

Triton 47 is the blog that Ajinkya kept on WordPress from May 2015 to August 2026 under the handle triton7777. This site is a re-structured, read-only archive of it: the same 544 posts, sorted into twelve threads, with an index note on each post and labels that say what is personal and what was generated by an AI assistant.

The original blog is still at triton7777.wordpress.com. Every post here links back to its original.

Who wrote it

Ajinkya came to Finland to study at Aalto University and stayed for a doctorate in a probability group, working on object tracking with Rao-Blackwellized particle filters, trajectory behaviour models and thermal imaging. The doctorate was completed in 2025. The one research project the blog describes in detail is Eyes of the North: ship-borne multispectral (RGB and infrared) segmentation of ice tracks for autonomous navigation in the Baltic, with a released dataset and a benchmark of 26 networks.

Outside research he builds things: two Bambu Lab printers, a 3D-printed Arctos robot arm (build day 1, first motion), XREAL glasses, a Meta Quest 3, a Raspberry Pi 5 with an M5Stack LLM8850 accelerator, small displays and microcontrollers. Code lives at github.com/ajinkyagorad and printable parts at GrabCAD.

He coined two words that recur across the blog: experi, the raw fact of experiencing, and consci, the thing that experiences (Recipe for everything).

How the blog was written

Almost everything was written in one pass, without planning or later edits, often on a phone, sometimes dictated. From 2023 the author started pasting in conversations with ChatGPT, Gemini, DeepSeek, Grok and Google's AI mode, usually saying so. By 2026 many posts are dialogues in which the author's own words are the short prompts and the assistant supplies the bulk.

That is why this archive labels authorship on every post:

• no label — the author's own writing;
• AI-assisted — the author's writing combined with pasted AI text;
• AI-generated — mostly AI output that the author prompted and kept.

The labels are editorial estimates made while reading each post. They can be wrong in either direction; the original post is one click away.

What the editorial pass did

• Read every post, the About page and the comments.
• Assigned each post to one of twelve threads. The original blog had a single category ("Uncategorized") and tags only from December 2023 onward.
• Wrote a one-line index note for each post, shown as Index note above the text. The notes are a reader's shorthand, not a summary the author approved.
• Marked posts that are personal entries. They stay in the archive as part of the record, but they are not surfaced in picks or related-post suggestions, and search engines are asked not to index them.
• Marked posts that contain a project idea and consolidated those ideas on the Ideas page.
• Collected the recurring theses on the Canon page.
• Dropped the WordPress comment system. In eleven years the blog received one comment from a reader; the rest were the author's own cross-links, which now appear as related posts.
• Left the texts unchanged, including typos, dictation artefacts and unfinished drafts. Empty posts are kept and marked.

The magazine

A printable magazine was made from the archive in September 2026: 108 pages, A4, with 53 posts, the canon, the maker photographs, the sketchbook pages and the ideas list. Two files: the reading version (cream paper baked in, for screens and home printing) and the print-shop version (white paper, 3 mm bleed, crop marks).

Technical notes

The archive is a static site generated from the public WordPress API dump of the blog (posts, tags, comments). Images hosted on WordPress were mirrored; a few 2018 images that lived on Slack's CDN are gone and are marked as missing. Old post URLs of the form /YYYY/MM/DD/slug/ redirect to the new pages. There is an RSS feed and a client-side search.

Numbers about the corpus are on the Stats page.