Amazon Web Services (AWS)
AI Prompt Engineer — AWS Bedrock (ML Data Associate II)
Chennai, Tamil Nadu, India
- Designed and optimized 50+ structured prompts for summarization, classification, reasoning, and conversational AI — boosting AI pipeline response accuracy by 30% across real-world production settings.
- Evaluated 2,000+ complex LLM outputs against strict quality benchmarks, grading relevance, factual alignment, and instruction-adherence; fed model gaps and edge cases directly to engineering teams.
- Conducted adversarial red-teaming campaigns to detect prompt vulnerabilities, biases, and safety failures — documenting actionable feedback to enforce Responsible AI compliance cross-functionally.
- Designed agent workflow prompt rules: precise structural instructions, contextual bounds, and JSON schema outputs for repeatable task execution.
- Vetted and benchmarked chatbot behaviors with advanced testing workflows (AWS Q) to decrease prompt response variance at scale.
Impact & milestones
+30% AI response accuracy in production pipelines2,000+ LLM outputs audited with zero-variance logging100% adherence to standardized multi-agent output structures
Prompt EngineeringLLM Output EvaluationAmazon BedrockResponsible AIRed TeamingPythonRAG Workflows
