tom/neuromancer
1
0
Fork 0
Code Pull requests Packages 2 Activity Actions
Self-hosted, GPU-optimised GenAI platform providing a drop-in OpenAI-compatible API
21 commits 2 branches 0 tags 1.4 MiB
Find a file
Exact
Tom Foster 8887d7e449 ci(vllm): make build workflow manual-dispatch only 
Remove automatic triggers (schedule and push) to ensure build parameters
are explicitly specified via workflow dispatch, consistent with other
CUDA/PyTorch workflows that require manual configuration.
2025-09-15 15:10:29 +01:00
.forgejo/workflows ci(vllm): make build workflow manual-dispatch only 2025-09-15 15:10:29 +01:00
docker feat(vllm): orchestrate dependency builds with prebuilt wheel assembly 2025-09-15 14:43:45 +01:00
docs feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
src/neuromancer feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
templates feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
tests feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
.gitignore feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
.markdownlint.yaml feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
.pre-commit-config.yaml feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
config.yml feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
CONTRIBUTING.md feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
docker-compose.yml feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
LICENSE feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
pyproject.toml feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
README.md feat(platform): establish Neuromancer AI orchestration platform foundation 2025-09-05 17:21:34 +01:00
uv.lock feat(ci): add comprehensive CUDA package wheel building infrastructure 2025-09-15 12:53:30 +01:00

README.md

Neuromancer AI Orchestrator

A self-contained AI orchestrator that provides OpenAI-compatible APIs for locally hosted models. Run voice synthesis, chat completions, and tool-enabled AI services through Docker containers with automatic resource management and zero-configuration model downloads.

Point any OpenAI-compatible application at this server and it handles the rest - no code changes needed.

What it does

Neuromancer orchestrates AI services through Docker containers, providing a unified OpenAI-compatible API for different types of models. The system handles container lifecycle, GPU memory allocation, and automatic cleanup - services start on demand, share resources efficiently, and restart automatically if something crashes.

Language models are powered by vLLM and support chat completions, tool calling, and agent workflows. You can configure different models for different purposes: lightweight chat, tool-enabled assistants, or specialised agents for complex tasks.

Voice services use Kyutai Moshi to provide zero-shot voice cloning, real-time speech-to-text, and text-to-speech. Generate voice embeddings from single audio samples and synthesise speech in real-time using your own voice or reference samples.

Documentation

Setup and deployment:

Using the API:

Technical details:

Ethical use

Voice cloning technology should be used responsibly:

  • Use your own voice or obtain explicit permission
  • Be transparent about AI-generated content
  • Consider the impact on voice actors and content creators
  • Do not impersonate others without consent or create misleading content

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.