Meghanath kethireddy

About

PayPal ML/AI practitioner who built and productionized a hybrid recommendation engine (BERT/LLM embeddings + collaborative filtering + XGBoost ranking) on AWS with end-to-end MLOps and orchestration. Addressed real-world issues like cold start and embedding latency (ONNX, clustering, caching, PySpark/Delta Lake) and drove a 27% lift in upsell conversion via A/B testing and stakeholder collaboration with marketing.

Experience

Education

Key Strengths

• Built and deployed a hybrid recommendation engine (BERT + collaborative filtering + XGBoost ranking) to production on AWS
• Delivered measurable business impact: increased PayPal upsell conversion by 27%
• Solved key production challenges: embedding latency, cold-start users, and scalable data pipelines (ONNX, PySpark/Delta Lake, caching)
• Strong MLOps/orchestration experience (Airflow DAGs, AWS Step Functions, SageMaker deployments; Prefect and Kubernetes CronJobs)
• Rigorous evaluation and reliability approach for AI workflows (offline + adversarial evals, shadow deploys, A/B tests, monitoring/guardrails/fallbacks)
• Pragmatic model/retrieval/prompt selection based on constraints (latency, accuracy, context length, data availability)