SealCheck
A tool I’m building to assess sandbox security in preflight mode, before AI models are tested inside it.
A focus on sandbox security before model testing begins.
A personal systems lab
New York, NYI’m Rakshan. A security engineer with a builder’s itch. This is where I share the systems I’m building, the decisions behind them, and what I’m learning along the way.
Currently building SealCheck
Sandbox security, before the model enters.
In development
Before the model enters.
SealCheck
I’m building SealCheck to assess sandbox security before model testing begins. The specific checks and implementation are still in development.
01 / Software02 / Artificial intelligence03 / Cybersecurity
From production security platforms to experiments on my own machine.
A tool I’m building to assess sandbox security in preflight mode, before AI models are tested inside it.
A focus on sandbox security before model testing begins.
A self-hosted coding system with builder and reviewer agents, human approval, and deliberately limited access to tools and credentials.
Isolated execution. Human approval. Credentials kept outside containers.
A vulnerability management platform that turns scan results into clear priorities, remediation deadlines, and Board reporting.
Post-scan triage reduced from weeks to minutes.
A threat-intelligence service that connects public feeds with asset inventory to surface the threats that need attention.
700 daily records narrowed to 1–5 actionable items.
Currently exploring
Agent orchestration. Sandbox isolation. The interesting space between giving a model useful tools and deciding what it should be able to do.
Inside my agent harnessAway from the day job
Custom gaming rigs, a homelab that is never quite finished, and rabbit holes in physics and quantum mechanics. I like understanding how the pieces fit.
The person behind the projects