Skip to content
Apart Research
Apart Research
Contact us
Sprint projectMay 25, 2026Los Angeles

NeuroTrace: Spec-Aware Neural Network Runtime

Robert Joseph George · Team Safeboy

Submitted to The Secure Program Synthesis Hackathon. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: NeuroTrace: Spec-Aware Neural Network Runtime

Open in a new tab (opens in new tab)
Recording (opens in new tab)Code (opens in new tab)
Share

NeuroTrace is a spec-driven observability framework for neural network models. The core idea is that a neural specification should not stay as a static file: it should become a live contract that the runtime, training loop, optimized implementation, export artifacts, and verifier outputs must continue to satisfy. NeuroTrace triangulates across TorchLean-style specs, PyTorch hooks, torch.fx, ONNX/VNN-LIB, Marabou, SpecTrace training logs, Fast Path checks, and SpecFuzz mutations to detect when the artifact chain drifts from the intended model. My long-term goal for it is to become a practical “spec observability” layer for ML systems, especially as AI agents generate code and frontier-style systems replace readable models with faster optimized runtimes.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Your private critique is not published; only the Apart team reads it. If you agree below, we share your review with grantmaking.ai (opens in new tab) and the Transformative AI Fund so strong projects can be funded.

Not shown on this page.

Shown on this page, without your name.

Only the Apart team reads this, and funders if you agree below.

Share my name publicly on grantmaking.ai *
Share my private critique with funders *

How much would this matter for AI safety if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical AI safety problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.

  1. This approach shows promise and is I believe a novel combination and application of these methods. I am excited to see if this scales to real-world bugs and implementation. I also liked the variety of objects covered in the demonstration here and was impressed by the hackathon velocity.

    I'd have liked to see more justification or exploration that the autoformalization into the contract is reasonable, since the correctness of this is fairly load bearing for the usefulness of this technique. Additionally, the write-up was a little hard to follow and some key claims or ideas were buried in the stream-of-consciousness format.

  2. Constant surveillance as opposed to a single check seems very promising. But, the human prompt to contract writing using an AI in itself is not a very robust methodology.

Cite this project

@misc{george2026neurotrace,
  title = {{NeuroTrace: Spec-Aware Neural Network Runtime}},
  author = {Robert Joseph George},
  year = {2026},
  month = may,
  note = {Submitted to The Secure Program Synthesis Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/neurotrace-specaware-neural-network-runtime-mipv}},
  url = {https://apartresearch.com/sprints/projects/neurotrace-specaware-neural-network-runtime-mipv}
}

More from The Secure Program Synthesis Hackathon

All 61 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026