Hi Hugging Face team,
I’m building ARIA (AI-Powered Repository Intelligence Agent), an open-source developer platform designed to help engineers understand, investigate, and safely modify unfamiliar software repositories.
Project
License: MIT
ARIA combines static code intelligence, semantic retrieval, execution analysis, and AI-assisted investigation into a single developer workspace.
It currently provides:
-
Repository-wide AST analysis using Tree-sitter
-
Symbol indexing and code intelligence
-
Dependency and call graphs
-
Architecture intelligence
-
API surface and contract analysis
-
Semantic code retrieval and RAG
-
AI-assisted repository investigation
-
VS Code integration
-
MCP integration
-
Docker-based self-hosting Why compute support is needed
ARIA’s analysis pipeline is significantly more computationally demanding than a typical web application.
For each repository, the system may perform: Git acquisition → parsing → symbol extraction → graph construction → embedding generation → vector indexing → repository intelligence generation
The hosted service therefore needs enough CPU and memory to perform repository-wide analysis reliably while supporting concurrent developer requests.
The project is already containerized and designed to run as a production FastAPI service, so I am not requesting infrastructure for an experimental prototype. I am looking for compute support to provide a public, limited demonstration environment for an already-developed open-source project.
Why Hugging Face
I would like to host the public ARIA demonstration on Hugging Face Spaces because Hugging Face provides an excellent environment for making open-source AI developer tools accessible to the community.
At the moment, my account shows Docker Spaces as requiring a paid plan. As an independent student/open-source developer, I am trying to maintain ARIA without personally taking on recurring infrastructure costs.
I would therefore like to ask:
Would ARIA be eligible for a Hugging Face Community Hardware Grant or another form of open-source compute sponsorship that would allow me to run the existing Dockerized application on suitable CPU hardware?
I do not require a GPU specifically. CPU-based compute with sufficient memory for repository analysis would be enough.
Current project maturity
ARIA already includes:
-
Production-oriented Docker deployment
-
Concurrency-safe repository acquisition
-
Branch-isolated analysis targets
-
Bounded background analysis workers
-
Retrieval performance optimizations
-
Qdrant vector database integration
-
Gemini → DeepSeek provider failover
-
MCP API boundary
-
VS Code integration
-
Automated backend and frontend test suites
-
Self-hosted deployment support The hosted instance would be intentionally constrained with limits on:
-
Repository size
-
Number of files analyzed
-
Concurrent analyses
-
Analyses per user
-
Chat usage
-
LLM token consumption
-
Maximum analysis duration
This would allow the hosted environment to remain a controlled community demonstration rather than an unrestricted public service.
Goal
My goal is to make ARIA useful to the open-source developer community while keeping the project completely self-hostable.
A community-supported Hugging Face deployment would allow developers to try ARIA immediately without installing the full analysis stack themselves, while the MIT-licensed repository would remain available for anyone who wants to run it independently.
Thank you for considering the project and for supporting open-source developers.
I would be grateful for any guidance on the appropriate Community Hardware Grant or sponsorship process.
Best regards,
Varshith Reddy