Developer Replaces Vector DBs for AI Memory with Git and Markdown

·2 views

A software architect has released an open-source AI memory system that uses Git and Markdown instead of vector databases or graph frameworks. The developer argued that spinning up heavy infrastructure just to store conversation context is unnecessary over-engineering for many use cases. The system leverages Git for tracking and rolling back an AI's conversation history, while Markdown serves as a lightweight, human-readable storage format that LLMs can parse without embedding pipelines. The solution is designed to be serverless, zero-cost, and portable enough to integrate into internal automation tools without database maintenance overhead. The project has been published on GitHub, and the developer is inviting the community to test the architecture and contribute feedback or pull requests.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Japanese Court Invalidates RED's RAW Video Patent in Landmark Ruling

A Japanese court has overturned a patent held by RED related to RAW video recording technology. The ruling marks a significant legal defeat for RED, a company known for aggressively defending its video compression patents. Panasonic succeeded in challenging the patent where major tech and camera firms like Apple, Sony, and Nikon had previously failed. The decision could have broader implications for the camera and video industry, potentially opening up RAW video technology to wider use by manufacturers.

0 comments Read more at Hacker News

ProgrammingDEV Community ·

Free browser tool converts cron expressions into plain English with live previews

A developer has built a free, client-side cron expression generator to help users who struggle to remember cron syntax. The tool lets users select day, hour, and minute values from dropdowns and instantly translates the resulting expression into plain English. It also previews the next five scheduled runs in the user's local timezone, helping catch daylight saving time errors. Ready-to-use code snippets are generated for Python, Node.js, Bash, Docker, GitHub Actions, and n8n. The tool requires no signup and runs entirely in the browser at cron-generator-kappa.vercel.app.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer finds three data leakage flaws in ML model that appeared to validate cleanly

A developer building a computer vision model to estimate container fill levels discovered three separate data leakage issues after the model appeared to perform well, with a mean absolute error of 0.055. Despite having grouped train/test split guardrails in place from the start, leakage occurred because a grouping column contained incorrect scene IDs instead of unique container IDs, meaning the split was not separating data as intended. A second leak arose when evaluation scripts bypassed the safe split module entirely and used frame-level rather than container-level grouping. A third issue involved a cross-site validation claim that broke down because a drone's default filename prefix appeared across multiple sites, causing training and test data to intermingle. The developer noted that while one published validation claim was invalidated, relative performance comparisons between model approaches remained reliable since all experiments ran on the same flawed partitions.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Documents Multi-Layer Debugging Journey to Connect PySpark with Kafka

A developer working on a financial data engineering project attempted to extend a batch pipeline with Spark Structured Streaming by reading JSON messages from a Kafka topic. Although the Python Kafka producer and Kafka broker were functioning correctly in Docker, running the Spark job triggered a chain of failures across multiple layers including PySpark, the Spark runtime, Kafka connectors, Hadoop on Windows, Docker networking, and dependency resolution. The spark-submit launcher failed to locate the pip-installed Spark environment on Windows, prompting the developer to run the Python script directly instead. This workaround bypassed the launcher issue but immediately revealed a second problem: Spark lacked the Kafka data source connector entirely, meaning it could not even attempt a connection. The experience highlighted the importance of isolating failures by layer rather than assuming all errors originate in application-level Python code.

0 comments Read more at DEV Community