Megan Wallace

COMPETENCY PROFILE

EXAMINER’S NOTE

You demonstrated exceptional strength in getting real work done and in checking the work, and both showed up concretely across your scenarios. Your model-selection reasoning was consistently sound: you matched tiers to the job by weighing volume, error cost, and latency rather than reaching for brand names, reserving top-capability models for high-stakes strategy and contract interpretation while using lighter models for triage and mechanical formatting, and pairing purpose-built extraction with a capable-model fallback. Your verification instinct is a clear signature — you placed review immediately after extraction because downstream systems trust structured output and errors compound silently, and in the survey task you insisted on themes backed by counts and distributions while refusing to force-fit ambiguous comments. Your incident-handling judgment was equally role-aware: you contained spread, distinguished deletion from data being permanently gone, and escalated to manager and legal rather than quietly fixing the problem. Your prompt design also showed real craft, with explicit format constraints, scoping guardrails like "flag rather than guess," and a discerning sense of which judgment tasks — cross-clause risk and deal-specific context — must stay with a human.