Sovereign AI evaluation with Sferical AI

Problem

Sensitive enterprise workloads require a deliberate comparison of model quality, reliability, latency, cost, hosting, access, and operating responsibility.

Contribution

Built a local evaluation-harness foundation and compared open-model and cloud options, including GPU/runtime choices and ownership requirements.

Design decision

Two-model operation and measured savings remain objectives, so the work is described as evaluation/planning rather than a completed production rollout.

Outcome and evidence

The evaluation is ongoing. No measured cost benefit or broad adoption is claimed.

Source: public CV online or dated CV (PDF). No private project documents are linked.