Swetlina Chanda

I am an AI and LLM Evaluation Specialist with 3+ years of hands-on experience in multilingual data annotation, AI response quality testing, and search relevance optimization. I have driven measurable improvements in model reliability and safety across Alexa LLM, Prime Video, and internal Generative AI systems by identifying hallucinations, refining prompts, and optimizing annotation guidelines.

I lead data curation and evaluation initiatives that improved annotation accuracy and model relevance, including projects that increased training quality by 18% and streamlined feedback triage for engineering teams. I collaborate closely with linguists, data scientists, and cross-functional stakeholders to ensure culturally appropriate, privacy-conscious, and high-quality AI outputs.