I want to explore how effective RAG-focused LLMs can be in accurately answering complex or nuanced questions about improving sophisticated systems. With over a decade of experience as a software architect, I have formed some opinions about what constitutes a good signal when analyzing large amounts of code for this purpose. So, I wrote this web application, an MVP that allows you to ask a question (once a day) about the software architecture of a popular open-source project. You can ask about the current architecture or about how to enhance that architecture to accommodate new capabilities or to pay down specific forms of tech debt.
Here is the list of open-source projects that have already been indexed: Cassandra, Debezium, Druid, Elastic Search, Lucene, Kafka, Neo4j, and Spark. Are there other open-source projects you would instead ask software architecture questions about? Feel free to post a comment about them or upvote another comment that already lists them.
The pipeline behind this app is compute-intensive. This project is self-funded and on a considerably lean budget. Waiting for the answer to your question would not be a good experience. There is no signup, but I do require an email address to verify that the question comes from you and to receive the answer. Outside of that, I have no interest in your email, so feel free to use a temporary email service such as Mailinator if you feel uncomfortable with providing your real email address.
This is a learning adventure for me, so I look forward to your feedback, especially regarding any hallucinations in the answer to your question. The app also has a rating and feedback collection experience, which you can use if you would prefer a little more privacy.
No comments yet.
