Skip to Content
Introduction

Hephaes

Inventory

Hephaes is a local-first open-source robotics log indexing and dataset conversion stack, built to turn raw ROS and MCAP logs into clean, searchable, reproducible datasets on your own machine.

Core Workflow

  • register local .bag and .mcap logs by file path or directory scan
  • index logs to extract duration, start and end time, topic summaries, message counts, sensor types, and raw metadata
  • inspect source assets and draft conversion specs from real data
  • preview drafts, confirm them, and save reusable configs
  • convert selected logs to Parquet or TFRecord with custom mappings, compression, resampling, and manifests
  • browse generated outputs and open artifact content directly
  • track durable jobs for indexing and conversion
  • review dashboard summaries, trends, and blockers for the local workflow

Open Source Package Today

The hephaes Python package is now independently usable as a local CLI and library.

It currently supports:

  • local .hephaes workspaces rooted in your project directory
  • path-backed asset registration without copying raw log files
  • a draft-first authoring workflow through hephaes drafts ... and hephaes drafts wizard
  • reusable saved conversion configs and later conversion runs from those configs

See the package docs here:

Enterprise Features

We are also building features for enterprise that sit on top of the local OSS core.

Planned enterprise features include:

  • cloud ingestion from buckets, remote URLs, and managed connectors
  • multi-user authentication, organizations, workspaces, roles, and ownership
  • shared catalogs with team-wide browsing and admin views
  • saved searches, shared presets, and richer metadata search
  • managed conversion jobs with retries, scheduling, and distributed execution
  • first-class named datasets with versioning, sharing, approvals, and publishing
  • dataset lineage with hashes, creators, schema governance, and audit history
  • remote replay and visualization with access control and collaboration
  • team workflows for outputs, approvals, integrations, and downstream compute actions

If you are interested in being a design partner, please reach out to hello@hephaes.ai

Repository Layout

  • frontend/: React + Vite UI with a Tauri desktop shell
  • backend/: FastAPI sidecar/service backed by the hephaes workspace
  • hephaes/: the shared Python package
  • docs/: this documentation site (Nextra)

Python Setup

Install both Python projects for local development from the repository root:

python -m pip install -r requirements.txt

Or install them individually:

python -m pip install -e "./hephaes[dev]"
python -m pip install -e "./backend[dev]"
Last updated on
Overview