Jack Lakkapragada
Nashville, TN
Bio
Updated 07/04/26 · Provided by member · VerifiedJack Lakkapragada — Senior Solutions Architect at Salesforce (18 years enterprise AI and CRM). Independent AI safety researcher: 8-part behavioral audit series on alignment faking across Claude 4.x and open-weight models, merged PR to UK AISI's inspect_evals framework, three Apart hackathon submissions. Core finding: frontier Claude models showed zero alignment faking compliance while Llama 3.3 70B complied at 80% under identical conditions.
Links
Updated 07/04/26 · Provided by member · Verified- Personal Website
- https://lvjr3383.substack.com/
Projects
Grants
Updated 07/04/26 · By grantmaking.aiNo grants recorded.