Gopichand Amaraneni

About

Healthcare-focused ML/LLM engineer who built a production hybrid RAG workflow to automate prior authorization by retrieving from medical guidelines/historical cases (FAISS) and generating grounded rationales for clinicians. Strong in operationalizing ML with Airflow/Kubeflow/MLflow on SageMaker, optimizing latency (ONNX/quantization/async), and reducing hallucinations via evidence-only prompting; also partnered closely with clinical ops to deploy a readmission prediction tool used in daily rounds.

Experience

Education

Key Strengths

• Built and deployed production RAG system for healthcare prior authorization automation
• Improved retrieval quality via chunk-size/overlap tuning and domain-specific embeddings
• Reduced LLM latency using ONNX Runtime, quantization, and async SageMaker calls
• Mitigated hallucinations with strict structured prompting and evidence-only responses
• End-to-end orchestration of data/ML workflows with Airflow, Kubeflow, and MLflow
• Reliability-focused evaluation approach (benchmarks, stress/edge-case testing, post-deploy monitoring)
• Effective collaboration with non-technical clinical stakeholders to define thresholds and usable outputs