| --- |
| title: README |
| emoji: 🟡 |
| colorFrom: gray |
| colorTo: yellow |
| sdk: static |
| pinned: false |
| --- |
| |
| # Setloop |
|
|
| ### We build the infrastructure that takes AI from the lab into production. |
|
|
| **Engineering consultancy · AI infrastructure products · Applied research** |
|
|
| [Website](https://setloop.io) · [Services](https://setloop.io/services) · [Setloop Lab](https://setloop.io/lab) · [GitHub](https://github.com/setloop-io) · [Contact](https://setloop.io/contact) |
|
|
| Setloop helps teams design, build and operate production AI systems. Our work spans private and sovereign AI, GPU platforms, inference, training, security, observability and cost control. |
|
|
| ## What we do |
|
|
| | Consultancy | Products | Setloop Lab | |
| | --- | --- | --- | |
| | Architecture reviews, implementation plans and engineering support for production AI. | Tools for operating GPU capacity, controlling AI spend, securing agents and automating operations. | Applied research into privacy and security risks in distributed AI. | |
|
|
| ## Products and platforms |
|
|
| - **[GPU Cloud](https://setloop.io/gpu-cloud-platform)** — secure GPU rentals, deployments, private clusters and distributed training. |
| - **[AI FinOps](https://setloop.io/finops)** — token-level cost visibility connected to product work and customer outcomes. |
| - **[LLMTrace](https://setloop.io/llmtrace)** — security, policy controls and full-fidelity observability for production AI agents. |
| - **[ProofSeam](https://setloop.io/proofseam)** — evidence-led security assessments for AI workflows and model privacy. |
| - **[AutoOps](https://setloop.io/autoops)** — governed autonomous SRE for AI infrastructure and Kubernetes platforms. |
| - **[Automatic RL Research](https://setloop.io/autoresearch-rl)** — bespoke closed-loop systems for long-running reinforcement-learning experiments. |
|
|
| ## Research and open artefacts |
|
|
| Setloop Lab investigates privacy and security in distributed AI. We publish papers, datasets and reproduction artefacts so that findings can be inspected and extended. |
|
|
| - **[Privacy Failure in Split-LLM Training: The Returned Gradient Nullifies the Decoys](https://huggingface.co/papers/2609.04382)** — the returned gradient exposes the real rows in every audited frame. [alphaXiv:2609.04382](https://www.alphaxiv.org/abs/2609.04382) |
| - **[What Gradients Add to Text Leakage in Split Language Models, Counted per Token and per Document](https://huggingface.co/papers/2610.04128)** — measured per word the gain is modest; measured per whole document it triples. [alphaXiv:2610.04128](https://www.alphaxiv.org/abs/2610.04128) · [reproduction package](https://huggingface.co/datasets/Setloop/What-Gradients-Add-to-Text-Leakage-in-Split-Language-Models-Counted-per-Token-and-per-Document) |
| - **[The LLM-Attacker: A Unified Red-Team Framework for Privacy Evaluation of Split and Distributed LLMs](https://huggingface.co/datasets/Setloop/The-LLM-Attacker-Unified-Red-Team-Framework-for-Privacy-Evaluation-of-Split-and-Distributed-LLMs/tree/v1.0)** — the evaluation framework behind both studies: combined manuscript, code and full evidence release (v1.0); arXiv preprint in submission. |
|
|
| ## Work with us |
|
|
| Planning AI infrastructure, evaluating a platform or moving an AI system into production? **[Tell us what you are building →](https://setloop.io/contact)** |
|
|
