星帆逐影

@AstroVelaAI

A high-performance, multimodal-native engine for AI workloads.

vane.astrovela.ai

Joined June 2026

8 Following

39 Followers

星帆逐影
@AstroVelaAI
Nov 9
Vane Data 0.3.0 is out: run multimodal pipelines locally without Ray, use embedding functions in SQL and Python, and install on Windows, macOS, or Linux. Vane Context is now in private beta.
vane.astrovela.ai
Vane Data v0.3.0 | Vane
Local Runner, multimodal performance, embedding functions, cross-platform wheels, and Vane Context private beta.
星帆逐影
@AstroVelaAI
Oct 29
Vane turns recordings and images into text and vectors. Apache Doris (@doris_apache) connects them with orders, tickets and production batches for semantic search, traceability and quality analysis.
Apache Doris + Vane: Bringing Multimodal Files into AI Search and Analytics | Vane
From vane.astrovela.ai
星帆逐影
@AstroVelaAI
Oct 23
When video reviews seem to disagree, check the test conditions. Vane Data extracts claims from subtitles and frames. HawDB (@NowledgeMem) preserves evidence and source versions. See how we trace results back to the original clips.
Vane Data + HawDB: Building a Traceable Multi-Video Comparison Workflow | Vane
From vane.astrovela.ai
星帆逐影
@AstroVelaAI
Sep 28
Calling a model is one step. Parsing files, managing batches, and joining results require a pipeline. Vane Data brings SQL, Python, and AI into one data flow for PDFs, images, audio, and video. Why a multimodal-native engine?
vane.astrovela.ai
Why We Need a Multimodal-Native Engine | Vane
Why multimodal pipelines need a multimodal-native engine, and how Vane Data unifies SQL, Python UDFs, and AI Functions into one dataflow as simple as warehouse ETL.
星帆逐影
@AstroVelaAI
Sep 24
From a few PDFs to thousands, batching alone isn’t enough. Vane Data brings CPU parsing, GPU embeddings, and Milvus writes into one pipeline—so stages can overlap instead of waiting on each other.
From a Few PDFs to Thousands: Can Your Embedding Pipeline Keep Up? | Vane
From vane.astrovela.ai

星帆逐影

星帆逐影’s posts