A behind-the-scenes dive into a project at bitcrowd: how we broke down the planning, implementation and release of Charger, the replacement billing service of our client Steady Media.
Back in 2021, we helped our friends at Steady Media to design, plan and build an in-house alternative billing system that they could migrate their users to. The migration of their 600k+ users to the new system had to be seamless and preserve data integrity.
On a recent client project, redesigning a large Rails app, we found ourselves having to get Rails, Hotwire's Turbo and React on Rails to play together nicely. Each of the tools comes with its own amounts of magic™, so integrating them turned out to be a bit of a juggle.
Google took down the flag, what now? How to load a browser extension in Chrome for development in the post "--load-extension"-era.
With logits processing, you can force open source LLM to comply with a schema
you provide. Grammar Constrained Decoding saves you the process of
resubmitting malformed LLM responses. Here we show the process with Bumblebee,
and explain the background. Want a specific JSON format reliably? Read on!
Ever tried to get a poem from an LLM doesn't contain the letter `e`? Large
Language Models are kind of amazing and surprisingly unreliable at the same
time. Using Elixir's Bumblebee and open source LLMs, you can get much better
control over the generation. In this article, we introduce you to logits
processing, and how it can be used to achieve what ChatGPT can not accomplish.
Find out how!
Your search works… but is it actually good? In our previous post, we showed why full-text search struggles with the long, natural-language queries common in MCP interaction. In this follow-up, we dive into dataset generation, evaluation, and search improvement. How do you know if a new embedding model is really better?"
Most text search solutions are fine tuned to serve results for keyword style queries. This is a problem in the age of MCP, as comparably long user input hits the same old pg_trgm search
This time, bitcrowd (almost) takes a break from AI and focuses on developer happiness. Discover how you can speed up your side tasks and read this before your AI sidekick notices.
We take our blog post series on RAG to the next level: Meet exmeralda.chat, an online demo for Elixir-based RAG systems. - And a quiet useful tool, too!
