PRASAD K

Designed and deployed a production RAG platform integrating large language models for healthcare prior authorization workflows, enabling grounded retrieval from clinical policy and drug formulary data.

Architected LangGraph-based multi-agent systems to automate eligibility verification, formulary lookups, and coding validation at scale, backed by secure Python microservices with PHI compliance.

Led optimization efforts that reduced token consumption and latency by 35%, and established evaluation pipelines that improved grounded-answer quality by 20% across benchmarked sets.

Built and maintained ML serving infrastructure for real-time fraud detection and risk scoring, supporting high-throughput transaction analytics at fintech scale.

Developed end-to-end pipelines for model training, evaluation, deployment, and automated retraining — migrating legacy models to cloud-native infrastructure with a 35% reduction in prediction latency.

Collaborated with compliance and audit stakeholders to implement model explainability frameworks and maintain governance artifacts aligned to regulatory requirements.

Built predictive models and NLP pipelines for customer analytics — covering churn prediction, multilingual sentiment analysis, and subscriber segmentation across large telecom datasets.

Designed scalable feature engineering pipelines processing petabyte-scale network and usage data, reducing data preparation time significantly via distributed PySpark jobs.

Deployed ML models as REST APIs and maintained interactive dashboards surfacing subscriber health metrics and NLP-derived sentiment trends for weekly stakeholder reviews.