Top ideas in the MariaDB-Python hackathon

This is part of a blog series on the MariaDB-Python Bengaluru Hackathon announced in August 2025. Want to organize a hackathon with MariaDB? Read more about MariaDB hackathons.

As we blogged last week, the ideation phase of the MariaDB-Python hackathon ended with a heap of ideas. Due to the sheer mass, we had a tough time evaluating suitable ideas and teams for the development phase.

Now that we have proceeded to the development phase, we are really looking forward to seeing all short-listed ideas materialize to code! Let’s take a look at a few ideas from each track that stood out of the crowd.

A short recap: the hackathon is about Making MariaDB easier to use by two approaches, reflected by the two tracks: Integrations and Innovations. 

Ideas in the Integration track 

Integrations making MariaDB easy and convenient to use from existing frameworks, tools, and environments. We are expecting the end result to be code in the form of pull requests into existing GitHub projects, or separate “connectors” that provide the key for users to use MariaDB in whatever setting they need a database.

We shortlisted 50 integration ideas from over 100. Let’s take a look. 

A pick of interesting integration suggestions to existing projects: 

• Dagster, orchestration platform for data assets, 14 k stars on Github 
• Haystack, AI framework, 23 k stars
• Django, vector support for the CMS, 84 k stars
• Autogen (and LangGraph), agentic AI, 51k stars
• MindSQL, vector integration, “Text-to-SQL RAG Library simplifying database interactions”, 423 stars
• Open WebUI, User-friendly AI Interface , 112 k stars
• Feast, Feature Store for AI/ML, 6 k stars
• Apache Airflow, platform for workflows, 42 k stars

For a successful contribution, we recommend contacting the target project early on to share your approach and get feedback on it. The suitable integration approach can vary between making a totally dedicated MariaDB integration, to making sure existing MySQL integrations are documented and tested for MariaDB compatibility. 

There were a couple of next level python connector improvement suggestions. Please be in contact if you need MariaDB assistance in these:  

• “Async MariaDB Connector for Python”, an improved python connector
• “mariadb-vector-python”, “a comprehensive Python library for MariaDB Vector operations, inspired by pgvector-python, but tailored for MariaDB’s unique capabilities.”

There were several idea submissions mentioning MariaDB Jupyter. We tried to accept those that we felt would be closest to a plausible addition to the mariadb_kernel, in other words a pull request adding magics for MariaDB Vector and LLMs that would make it easy to showcase how to do RAG with MariaDB.

There were a few “natural language to SQL query” suggestions that wanted to implement their solution from scratch.. We accepted a couple of these, but we recommend considering finding existing “text2sql” solutions where MariaDB could be added. 

We had to drop idea submissions for a variety of reasons. Some had generic bundling ideas around ML and AI tools with MariaDB. Since the criteria in this track was to submit a specific integration to an existing framework or project, these were deemed outside the scope for the integration track.

Some idea submissions were suggesting integrations for projects that already have MariaDB integrations, like LangChain. We hope we didn’t misunderstand the intent of such idea submissions, but we had limited time to look into each idea submission in more detail.

Ideas in the Innovation track 

Innovations make MariaDB easier to use by showing off code that uses cool MariaDB Server features. We are expecting the end result to be a  reference implementations in a GitHub repo that provides an ideal starting point for users to copy and paste from, to extend, to apply to their own situation.

To be honest, I wonder if we should have called the track “MariaDB feature use cases”, because we got so many innovative idea submissions that seem to be “cool apps” with the focus heavily on the end user and even business plans, when we had goal of focusing on technical features in MariaDB.

That said, what were the most popular MariaDB features to show case in the idea submissions? We short-listed nearly 100 exciting ideas that span MariaDB features such as:

• MariaDB Vector, a building block in AI, RAG and semantic search is unsurprisingly a popular choice in the idea submissions. An interesting twist in one submission was doing multimodal searches on both images and text. 
• ColumnStore – lightning-fast analytics queries on large datasets
• Temporal Tables – time-travel queries on what your data looked like in history
• Galera Cluster – MariaDB’s synchronous multi-master replication for high availability and zero-downtime deployments
• System-Versioned Tables – automatic tracking of all changes to your data with complete audit trails
• MariaDB’s Spatial/GIS capabilities – store and query geographic data with functions like distance calculations and location searches
• Virtual Columns – values are computed from other columns using expressions

We are looking forward to interesting code examples with step by step documentation explaining how to use the MariaDB features in the interesting use cases brainstormed for the track.

The idea submissions we have read through had some amazing, creative use cases, but we remind you, the focus is on how the chosen MariaDB feature can be showcased.

We highly recommend using the common, public dataset openflights to showcase these MariaDB features. The goal is to collect several showcases using that same dataset.

Heading into the hackathon’s development phase

As we continue in the development phase with a bit under three weeks to go, remember what has been mentioned: 

Everybody can stand out in the crowd, not just the prize pool winners. All meaningful contributions will get listed and thus get widespread attention from the MariaDB user community. 

Looking forward to your participation!