Joshua Rotimi
AI Output Evaluator & LLM Quality Specialists
New to Contra
Joshua is ready for their next project!
Evaluated 60 multi-turn conversations from a retail support AI to track context retention, persona consistency, and task completion across longer exchanges. Found that context drops clustered around the third and fourth turn, pointing to a memory window limitation rather than random failures, and flagged each instance with the exact turn where context broke down.
1
38
Ran 48 adversarial prompts against a production LLM to surface jailbreak vulnerabilities, role override attempts, and context manipulation before a public launch. Found 11 vulnerabilities including 3 critical, documented every failure with reproducible prompts, and delivered a prioritised risk report with fixes.
1
35
Built a custom evaluation rubric for scoring AI customer support outputs across five weighted dimensions: factual accuracy, reasoning validity, instruction adherence, adversarial robustness, and format compliance. Each dimension includes plain-language scoring criteria so any team member can apply it consistently, no AI background required.
0
34
A sample LLM evaluation report analyzing AI responses for accuracy, reasoning quality, hallucinations, and failure patterns. Demonstrates structured testing and quality assessment workflows.
1
140