Likely AI-generated. Pangram flags 90% of this text as AI-written.
2
Hypervisor-Zero
1
ProposalGrant
Closes November 22nd, 2026$0raised
$1,000minimum funding
$10,000funding goal
Offer to donate
10 daysleft to contribute
You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$10,000 left to reach the funding goal.
Project summary
I am a solo working on Hypervisor-Zero, an independent research project focused on AI control, agent security, and AI powered cyber defense, the project aims to solve an important problem (how do we make sure AI agents stay within safe boundaries when they use tools, access systems, and make decisions?)
I am building a system that can monitor agents, control their actions, detect unsafe behavior, and provide evidence of whether security controls actually work I have completed the initial foundation, including a working prototype, a controlled testing environment, and a research dashboard i am now moving into Phase 2, where I want to build specialist AI agents that can work together to detect threats, investigate suspicious activity, analyze security risks, and support cyber defense operations so to make this possible, i am preparing a small datasets and fine tuning models for security decision making, threat detection.
Request access to the Researcher dashboard at:
Or
info@hypervisorzero.xyz
What are this project's goals? How will you achieve them?
The goal is to deliver a practical research demonstrating how coordinated, controlled AI agents can improve cyber defense
How will you know it worked?
Success will be measured by threat detection accuracy, false positive rates, time to identify and contain threats, and the system’s ability to block unauthorized agent actions even when agents are compromised or manipulated...
How will this funding be used?
$3,000 for GPU rental, covering approximately 400 hours on an H100 at $2.89 per hour ($1,156) and 420 hours on an H200 at $4.39 per hour ($1,843.80), giving about 820 GPU hours for fine-tuning and testing.
$1,500 for API usage, which would cover approximately 75k model requests based on an estimated average of 10k input tokens and 2k output tokens per request using over different models
$1,500 for expert dataset review, based on two reviewers working 50 hours each at an estimated $12 per hour, plus 25 hours of independent adjudication at the same rate.
$1,800 for cloud hosting, secure testing environments and storage over 12 months, based on an allowance of $150 per month
$600 for software and development subscriptions over 12 months,
$1,000 is allocated for independent technical testing, estimated at 40 hours at $25 per hour, while the remaining $600 will cover unexpected project costs.
The total budget is $10,000.
Who is on your team? What's your track record on similar projects?
I am (Muhammad) a third-year software engineering student and independent researcher I'm currently solo and I previous AI safety work focused on developing agricultural datasets and fine-tuning models to understand agricultural knowledge and local language's
What are the most likely causes and outcomes if this project fails?
The likely outcomes would be delayed development, incomplete experiments, failure to demonstrate measurable results, or eventually stopping the project and even if it fails, at least the work can produce useful research,benchmark datasets, and lessons for future AI safety and systems.
How much money have you raised in the last 12 months, and from where?
I self funded my work using personal funds and limited resources I have not raised significant external funding yet!