Senior Software Developer in Test

Wand AI has integrated VLNO's model robustness technology into its sovereign AI platform, enabling governments and regulated enterprises to deploy open-weight models on their own infrastructure with certified security. VLNO continuously tests customer models against adversarial attacks and generates training data to strengthen them, with protections incorporated directly into the model weights. The integration addresses a key barrier to production deployment of open-weight models, which often lack standardised safety evaluations and proven adversarial robustness. Before models enter Wand's Open Model Registry, VLNO benchmarks them against adaptive attacks. Models failing to meet robustness thresholds receive adversarial training data that customers can incorporate into their fine-tuning pipelines. The joint reference architecture is now available for sovereign programmes and regulated enterprises. Wand AI, founded in 2023 and headquartered in Palo Alto, provides sovereign AI infrastructure for deploying AI agents.

Wand AI and Protopia AI have partnered to deliver sovereign AI infrastructure that allows governments to run sensitive workloads on shared computing resources without exposing data. The collaboration combines Wand's operating system for hybrid human-AI workforces with Protopia AI's Stained Glass Transform privacy technology. The solution addresses a critical national security challenge: currently, different government ministries and agencies must maintain separate infrastructure for sensitive data, resulting in massive underutilisation of computing resources. Protopia's technology converts sensitive inputs into protected representations at the trust boundary, allowing multiple agencies to share one sovereign backend whilst keeping their data secure. The joint architecture aims to increase national compute utilisation from roughly 25% to over 80%, reducing costs for participating organisations. The solution is available now to sovereign programmes and enterprises. Founded in 2023, Wand AI is headquartered in Palo Alto. Protopia AI is based in Austin, Texas.

Wand AI and Protopia AI partner to power sovereign AI on shared, secure, high-utilization backend. Wand AI's agentic workforce operating system paired with Protopia AI's Stained Glass inference privacy layer. Every ministry and agency in a nation can run sensitive workloads on a single sovereign AI back end, with plaintext never leaving the trust boundary. Wand AI and Protopia AI today announced a partnership to deliver sovereign AI at national scale. The collaboration combines Wand's operating system for the hybrid human and AI workforce with Protopia AI's inference privacy layer, enabling a government or enterprise to run its most sensitive workloads across one shared sovereign backend at full utilization, without exposing data to the infrastructure operator or to other tenants. Real sovereignty requires the entire AI stack to run on sovereign soil and sensitive data. The agentic layer can be sovereign, but the moment an agent sends that sensitive data to a model, that data is exposed in plaintext across the entire serving environment: logs, caches, GPU memory, traces, and runtime infrastructure. That creates a hard national-security problem. Secrets from different ministries, agencies, banks, and critical infrastructure operators cannot safely share the same backend, so the default today is fragmentation, a separate infrastructure stack for each data owner. The result is massive waste: national compute investments sit underutilized because every sensitive workload requires its own isolated environment. At country scale, this "isolation tax" can reach tens of billions in infrastructure spend over time. The alternative is worse: data that is sovereign in name, but not sovereign in fact. The Wand and Protopia AI partnership removes that tradeoff. Wand provides the sovereign agentic stack: country-level governance, a hybrid human and AI operating system, certified AI labor, and autonomous agents, attached to the sovereign compute base. Protopia AI provides the inference privacy layer. Its Stained Glass Transform(TM)(SGT), applied by the SafeCLAW subagent at the trust boundary, converts sensitive inputs into protected representations, so only the protected form ever crosses into the model and compute layers, with no decoder on the host and accuracy parity to plaintext. Because plaintext never leaves the trust boundary, every data owner can share one sovereign backend without exposing data to the operator or to one another. Too often, sovereign AI programs have had to fragment their agentic infrastructure into dedicated stacks whenever sensitive data enters the workflow. Wand and Protopia AI remove that constraint. Protopia Stained Glass keeps plaintext inside each data owner's trust boundary, so ministries and agencies can run their highest-value agentic workloads on one governed sovereign backend and make full use of the compute already in place.