From 56756b0d65f30dd4fc37c203b384a75af922f50f Mon Sep 17 00:00:00 2001 From: Mayur Kale Date: Thu, 12 Nov 2026 20:24:18 +0000 Subject: [PATCH] docs: add training data preparation to the Phase 1 roadmap Name training data preparation as a Phase 1 goal: file connectors for Parquet, JSON lines and Avro, a deterministic train and test split, and a tested Python example that reads a table at a pinned snapshot. The work is tracked in #116, #117 and #118. Unstructured files such as images, audio and PDFs are left out until the project can run Python or Wasm steps. Signed-off-by: Mayur Kale --- ROADMAP.md | 3 +++ 1 file changed, 3 insertions(+) diff --git a/ROADMAP.md b/ROADMAP.md index 7946399..5db4ce0 100644 --- a/ROADMAP.md +++ b/ROADMAP.md @@ -24,6 +24,9 @@ Goal: a small team can run it on their own cloud account. - Connector SDK v1: YAML connectors, Rust SDK, conformance tests - First connectors: Postgres, MySQL, S3 files, REST APIs, Kafka - Postgres wire protocol, so BI tools can connect +- Training data: file connectors for Parquet, JSON lines and Avro, a + deterministic train and test split, and a tested Python example that reads + a table at a pinned snapshot, so a model can name the exact data it saw - Web UI: project browser, run history, table preview - Compaction and snapshot expiry