Laxmi Biyyala

Architected an enterprise-scale agentic AI platform with 12+ autonomous agents for asset management, custody banking, portfolio analytics, and investment research. Built a Knowledge Graph RAG platform using Neo4j and GraphRAG for semantic search over 500K+ financial documents. Fine-tuned domain LLMs with PEFT/LoRA/RLHF to improve accuracy and reduce hallucinations. Implemented end-to-end LLM evaluation and AI observability (LangSmith, RAGAS, Arize AI, MLflow) to track factual accuracy, latency, and model drift. Developed predictive ML models for portfolio risk analytics and engineered AML/trade surveillance pipelines using PySpark, Kafka, Databricks, and Delta Lake. Established enterprise MLOps/LLMOps practices and deployed inference services using Docker, Kubernetes, OpenShift, Helm, and FastAPI, with enterprise AI governance controls (audit logging, RBAC/IAM, data lineage).

Designed and deployed end-to-end ML and Generative AI solutions for healthcare analytics on GCP Vertex AI. Built clinical NLP/NER pipelines using BioBERT/ClinicalBERT and Hugging Face Transformers to extract insights from 200K+ physician notes. Developed and fine-tuned domain-specific LLMs using PEFT/LoRA and RAG for clinical summarization and automated documentation workflows with LangChain agents. Built healthcare chatbots and AI assistants handling 10K+ monthly interactions. Optimized LLM inference on Vertex AI with batching, caching, and quantization to reduce latency and infrastructure costs.

Integrated FHIR and HL7 interoperability standards with EMR/EHR systems across 15+ hospital networks, reducing data integration and mapping errors. Engineered XGBoost/LightGBM/TensorFlow models for disease risk prediction, improving early cardiovascular detection across a multi-hospital network. Applied Explainable AI using SHAP and LIME to improve transparency for clinician recommendations. Designed MLOps pipelines with MLflow, Docker, Kubernetes, and CI/CD for model versioning, deployment, and monitoring, including automated model validation and A/B testing frameworks for production rollout.