2025-09-09 –, Room B
Search is a core and high user impact feature in many Django apps. However, queryset filters and keyword search in Django apps often fails to meet user expectations for search relevance, speed, and personalization.
This talk introduces vector search — a powerful new paradigm that goes beyond keywords to search, recommend, and analyze your data. You’ll learn the basics of vector embeddings, how vector search works, and how to integrate vector databases into your Django models using the django-semantic-search package — with practical code examples.
We'll show a demo of product search/recommendations in a Django e-commerce app, then briefly explore other use cases like multi-modal search, content discovery, clustering, and retrieval-augmented generation (RAG). Finally, we’ll cover the trade-offs of vector search to help you decide when it’s the right tool for the job.
Traditional search in Django apps, typically powered by queryset filters or keyword search, often fails to capture user intent. It matches exact words but misses meaning. Vector search solves this by representing the "meaning" of data (text, images, audio, video, etc) as high-dimensional vectors generated with ML models, enabling more relevant, personalized, and faster results.
In this talk, you'll learn:
- When traditional search with Postgres or Elasticsearch falls short
- What vectors are, how vector search works, and when vector databases help
- How to integrate vector databases into your Django models using django-semantic-search
- A demo of "More products like this" feature with vectors in a Django e-commerce app
- Brief understanding of use cases beyond text search: multi-modal search, recommendations, content discovery, clustering, anomaly detection, and retrieval-augmented generation (RAG)
- The trade-offs and limitations of vector search — when it helps and when it doesn't
Who Should Attend
This session is for intermediate Django developers familiar with models, views, and basic querying. Vector search is becoming a high demand skill so everyone is welcome. But if you've built search features with filters or keyword search and are curious about taking them to the next level with vectors, this talk is definitely for you!
Prerequisites
- Familiarity with Django ORM and basic database concepts
- Exposure to search implementation using Postgres or Elasticsearch is helpful but not required
- Basic understanding of machine learning is helpful but not required
Hi, I'm a Python and Rust engineer with a deep interest in search engines, AI, and open source. I started programming at 13 and have been working full-time as a software engineer for the past 3 years. I currently work for Qdrant, the most loved OSS vector database.
Django has been a core part of my development journey and I've used it extensively across projects and organizations. I'm passionate about building open-source developer tools and sharing knowledge through talks on topics that matter to me.
Based in Bangalore (India), I enjoy discovering global cultures and cuisines.