HN helped make this happen - I built a 0.1 version for one client (Snagajob). I encountered someone from Wikimedia Foundation on Hacker News who was going to build something similar. We ended up partnering after meeting on Hacker News, and now work together on version 1.0 of Elasticsearch Learning to Rank which ultimately has been used on Wikipedia's search[2].
Wikimedia Foundation ultimately had super deep Elasticsearch internal knowledge which helped tremendously, whereas I mostly worked on building out the docs after the initial prototype[3]. I'm rather proud of the docs as it's a bit of a mini book on Learning to Rank :)
[1] - Plugin itself http://github.com/o19s/elasticsearch-learning-to-rank
[2] - WMF article on the plugin https://blog.wikimedia.org/2017/10/17/elasticsearch-learning...
[3] - Docs http://elasticsearch-learning-to-rank.readthedocs.io
https://www.smart-edit.com/Writer/
A lesson I've learned is that coolness and commercial success are not tightly connected. You can build something great that users still will not be prepared to pay for.
B2B is the way to go, not B2C.
I regret not trying to market it more, or moving into the virtual cycling space -
https://dzone.com/articles/virtually-cycling-the-alps-with-a...
I had no idea what I was doing but it was super fun. Hopefully it's useful to some.
XgeneCloud now generates REST & GraphQL APIs on any SQL Database. Plus there is much more! :)
Supported databases : MySQL, PostgreSQL, SQL Server, SQLite, MariaDB & AWS Aurora
I made Artbuffer to create my own wall art without having to spend years learning to paint. I've successfully used it to create my own :-)
Some examples:
https://www.artbuffer.com/view?a=wslyHCoGjqoA3ZKgsQ_hNAlnshT...
https://www.artbuffer.com/view?a=-9Wr6aLtgF383zsAWxchXJbt8CW...
https://www.artbuffer.com/view?a=OuIs3I6vHAyqV-PoMb6EFiRVAFw...
Coolest computer related thing? I used to run http://uptime.openacs.org (which I port from Oracle -> Postgres) and http://myturl.com (a tinyurl clone written in AOLserver). But as things tend to do, both projects are shutdown.
I still find the cigar box guitars the cooler thing.
https://sixdegreesofkanyewest.com/
I built it back in 2016 and haven't updated it since so newer artists don't show up.
I also wrote up a concise guide on everything the extension checks for and why those checks are important: https://www.checkbot.io/guide/
I found working on the above solidified by existing knowledge I'd picked up as a web developer and filled in the gaps. It's satisfying to know the extension is helping people identify website problems every day and teaching them tips they didn't know before.
Also, I've been wary of what seems like SEO snake-oil advice in the past (like a lot of developers) so it was good to digest all the evidence-based SEO advice I could find and condense it into small actionable tips most developers would agree with.
It's open source: https://github.com/brettkromkamp/contextualise
It initially started off with a question of "Why is ES so difficult to grok and manage?". Then I started researching around various data structures for efficiently storing text and numerical values and various algorithms for sorting the data in real time to surface relevant documents.
After 4 years of development, the project is pretty stable but I'm still adding features and has been getting some traction finally. I recently added a raft based replication aka clustering. I used a production ready Raft library for that, but it was still very nice to understand the nitty gritties of the Raft implementation.
I doubt I would build something more complicated ever, but who knows right :)