Generalist AI Trainer (Contractor) — Abaka AI
Trained and evaluated AI model responses by applying project guidelines to produce precise annotations and feedback. Performed adversarial red-teaming style checks to identify weaknesses, biases, logical errors, and potential risks. Conducted fact-checking and research to verify information used in training data and model outputs. • Applied complex client instructions consistently across tasks • Compared different model outputs and documented constructive feedback • Identified model risks and edge cases for ongoing model improvement • Ensured adherence to safety and quality standards