dataaaaa!

Timestral provides standalone multivariate probabilistic time series forecasting with support for up to 32,768 steps and covariates. It runs inference on CPU or GPU with minimal dependencies like PyTorch, NumPy, and safetensors, generating 21 quantiles per time step across multiple series.

OLCF is an open-source lakehouse platform creating an Apache Spark Standalone cluster across custom machines, cloud VMs, or Kubernetes. It extends Microsoft Fabric by using olcf.run_remote() to offload notebook compute to owned Spark clusters while directly reading and writing OneLake.

DataForge is a Python and Streamlit tool for automated ML data prep. It provides an audit-trailed pipeline to clean, transform, encode, scale, and split datasets without rewriting pandas code, exporting model-ready data and reports.

Polish-sci-figures is an open-source pipeline transforming tabular inputs like CSV, TSV, and Excel into publication-ready visual assets. It retains data analysis records, effect estimates, and reproducible plotting recipes while supporting batch image processing and SVG, PNG, or TIFF exports.

pg_vectorsearch is a PostgreSQL extension for high-performance approximate nearest neighbor search. It introduces the PRISM index, using hierarchical centroid routing, RaBitQ quantization, and SIMD optimizations to outperform HNSW on 100M+ vector datasets with less memory.

Anima is a Claude Code skill running on Node.js 20.10+ that generates standalone HTML visualizations of biological mechanisms. Its drawing pipeline uses only Node built-ins with zero runtime dependencies, using Chrome solely for capture and verification scripts.

This repository indexes over 100 analytical datasets across domains like time use and crime. Sources include the American Time Use Survey, offering minute-by-minute activity logs linkable to the CPS, and standardized individual-level NIBRS crime records spanning 1991 to present.

Ferdio demonstrated how one simple dataset can yield 100 visual presentations by varying analytical properties

VisActor's VGraph is a high-performance graph visualization and analysis engine. Ideal for pipeline DAGs, mind maps, and network topologies, it supports smooth interaction of tens of thousands of nodes and edges, offering built-in node types and dynamic layouts for large graph data.

Schemagate provides text-to-SQL access control by filtering schemas based on caller permissions before prompting the LLM. By pruning unauthorized tables for engines like Postgres and MySQL, it prevents silent RLS failures and cuts prompt tokens by up to 75%.

QuantBot is a quantitative trading system built with Go, React, TypeScript, Vite, and Wails. Its architecture orchestrates an automated daily pipeline combining multi-factor screening, backtesting, real-time portfolio monitoring, audit logging, and rule-based risk management with LLM agent

Chart4D provides native Delphi charting for VCL and FireMonkey, outputting publication-ready data visualizations. Its core relies solely on the RTL with zero dependencies, rendering identical plots and PNG exports across frameworks using automated label placement.

dbt v2.0 rewrites the framework in Rust as a single binary, bypassing Python dependencies. It accelerates project parsing and compilation while generating Parquet metadata artifacts alongside legacy JSON to streamline transformation modeling in warehouses.

Prism enables sandboxed, offline visualizations in Obsidian notes, turning local CSVs and Markdown tables into reactive charts. It bundles tools like D3 and Chart.js into secure iframes without network access, allowing AI agents to generate code and inspect render snapshots via CLI.

SeeCode generates interactive diagrams for coding agents from JSON specs, supporting architectures like APIs writing to Postgres and Kafka. It exports to HTML, SVG, PNG, GIF, or MP4. Charts ingest CSV or JSON files directly so numerical data bypasses the model.

WorldCupROI provides a deterministic data and modeling pipeline that ingests match data, text signals, and fan behavior. It builds baseline models and prediction intervals for sports sponsorship analytics, outputting evaluation artifacts, scenario reports, and dashboard views.

ZON-Forge TS is a TypeScript-native tool that serializes Zonal Object Notation (ZON) documents into JSON. Designed for data pipelines, it transforms indentation-sensitive zone markers into standard JSON structures, providing a predictable transformation layer without requiring pipeline redesigns.

Dex is an agent-native analytics engineering toolkit for data warehouses and dbt projects. It automates transformation authoring, profiles schemas, and resolves schema drift while treating repos as the source of truth with reviewable diffs. It runs read-only against warehouses or DuckDB.

FlareDB is a Rust-based streaming database that uses Apache Beam as its interface. Its unified streams-and-tables architecture executes batch and streaming pipelines, letting PCollections transition between streams in motion and materialized table state within a single engine.

Parseable is a Rust-based, stateless columnar data lake for telemetry data. Built on open Parquet and object storage native architecture, it unifies logs, metrics, and traces while managing high-cardinality workloads at scale with SQL querying.

jev4pg adds natural language to SQL and semantic operators to PostgreSQL. It features 41 JEV operators for filtering, extraction, and verification. PostgreSQL manages joins and windows while JEV uses parallel execution, input deduplication, and evidence caching for efficient query planning.

Sqmeow.nvim integrates database management into Neovim, letting engineers browse schemas, execute queries, and stage edits directly in the editor. Powered by a Rust engine that pages large results, it supports PostgreSQL, DuckDB, ClickHouse, and exports datasets to CSV, JSON, or SQL INSERTs.

Awesomedataviz indexes 291 visualization and BI tools across Python, R, and JavaScript, tracking metrics via daily updates. Platforms like Grafana and Apache Superset offer SQL IDEs and metrics observability, while the catalog exposes an MCP server and JSON API for programmatic agent access.

Marimo Studio transforms reactive Python notebooks into interactive apps, dashboards, and static reports. It keeps computation, data, and controls centralized in the notebook while serving distinct frontend views. Notebook cells update dependently as input controls change.