Pixeltable @pixeltablehq

A new episode of Database School is out! I spoke with Marcel Kornacker, the creator of Apache Impala and co-creator of Apache Parquet, to talk about his latest project: @pixeltablehq, a multimodal database for the AI age. youtu.be/7nsfzb2bVpQ

Think of Pixeltable as a data infra specifically designed for AI applications that work with images, videos, audio, and documents. It's a database system that natively understands multimodal data and can orchestrate workloads. As a software engineer, you've probably dealt with…

Building an infinite-memory, context-aware, multimodal chatbot is as easy as this in Pixeltable: github.com/pixeltable/pix… Why? Because Pixeltable is the only multimodal data infrastructure that unifies storage and orchestration for your AI workloads. pip install pixeltable

Work with Videos, Frames, Model JSON Outputs, Leverage the power of @modal's GPUs right within their new notebook platform with Pixeltable! Build your evaluation pipeline, storage, cache, and orchestration for your media data. modal.com/notebooks/pbru…

Pierre Brunelle @pjlbrunelle

3 weeks ago

Work with Videos, Frames, Model JSON Outputs, Leverage the power of @modal's GPUs right within their new notebook platform with Pixeltable! Build your evaluation pipeline, storage, cache, and orchestration for your media data. modal.com/notebooks/pbru…

0 2 7 1K 5

Agents as Data-Centric Applications in @pixeltablehq 1) knowledge.add_embedding_index("content") 2) chat_history.order_by(timestamp).limit(4) 3) docs.similarity(query) 4) tools = pxt. tools(analyze, fetch, search) 5) agent.add_computed_column(answer=llm(all_layers)) 6)…

pip install pixeltable -> unified storage, orchestration, and retrieval for your multimodal ai workloads right there on your machine.

@eersnington @adamwathan @FFmpeg If you struggle with ffmpeg: x.com/pixeltablehq/s…

Pixeltable @pixeltablehq

4 weeks ago

@eersnington @adamwathan @FFmpeg If you struggle with ffmpeg: x.com/pixeltablehq/s…

0 1 3 387 2

Download Image

🚀 @pixeltablehq + @pydantic integration! Type safety in Pixeltable just got even better! Define your data models once, get validation everywhere across your storage and orchestration layers: -> pip install pixeltable pydantic | @samuel_colvin

Latest video-related built-in UDFs we added: clip(), extract_frame, segment_video(), concat_videos(), overlay_text(), and VideoSplitter!

If you're building AI applications today, you've probably experienced this pain: spending 80% of your time on data infrastructure and only 20% on the AI that actually matters. You're not alone. Traditional data engineering worked great for the SQL era. Build tables, run ETL…

𝗕𝗬𝗢 𝗲𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝘀 𝗶𝗻 𝗣𝗶𝘅𝗲𝗹𝘁𝗮𝗯𝗹𝗲. Pixeltable maintains embeddings on updates: • Return pxt.Array[(d,), pxt.Float] for exact dims • Cache the model inside the UDF • Use batch_size for throughput • Drop‑in add_embedding_index

@BuildWithJit @kurosci i'm building a lot on @pixeltablehq stack lately. Curious what you are using for your processing flow?

1h40 youtube video: Prompt Engineering in Python | Automatic and Programmatic Prompt Optimization | Complete Course How to code, your own, automatic prompt optimizer. How the most advanced prompt optimization tool, DSPy, works and how to fully leverage its capabilities. How…

Pixeltable SDK: the "primitives" to build real multimodal AI apps (really fast) - create_table: persistent, versioned memory for app state and media - create_view: iterator-driven chunking; incremental updates, no duplication - add_computed_column: declare DAG nodes;…

Your multimodal AI App just hit bad data. Now what?!?! 1) Your AI workloads should handle real-world, messy data without crashing. 2) You should be able to easily isolate and analyze failures without searching through massive log files. 3) You should be able to incrementally…

From One Video to Thousands of Searchable Frames: - Your 30-minute video became 1,800 searchable frames - Zero data duplication (frames materialized on-demand) - Cross-modal search (find frames using text descriptions and/or other images) - Automatic synchronization (add new…

An alternative to Text-to-SQL We launched pxt.retrieval_udf(), a feature that transforms any Pixeltable table into an AI-queryable tool for agentic workflows. Traditional RAG excels at unstructured data (at least Pixeltable does) but struggles with structured information. Now…

@pixeltablehq automatically versions every change to your data and schema - no configuration required. This isn't just backup functionality; it's a core architectural feature that enables: 1. Automatic Version Tracking - Every table operation (inserts, updates, deletes, schema…

Why juggle 5+ tools for your multimodal AI Apps stack when you can use 1.

@anshnanda_ Check out @pixeltablehq for your multimodal data infra needs!