Sai Krishna CH

Deployed enterprise-grade GenAI solutions supporting customer service, compliance automation, and fraud investigation workflows across regulated banking environments. Designed scalable Retrieval-Augmented Generation (RAG) pipelines indexing 5M+ internal documents with FAISS and Pinecone; implemented a hybrid BM25 + dense retrieval strategy to boost precision@5 by ~28%. Fine-tuned domain LLMs (GPT-4, Claude, LLaMA) using LoRA/QLoRA and PEFT to improve factual accuracy. Built multi-agent conversational workflows with context preservation and guardrails. Implemented monitoring with SageMaker Model Monitor and CloudWatch; established audit traces and model governance. Optimized inference costs via batching and cache strategies and developed reporting dashboards for operators and auditors.

Designed and delivered NLP/document intelligence solutions for knowledge automation. Built semantic search systems with Sentence Transformers, FAISS/Pinecone, and knowledge graphs (Neo4j) to improve retrieval and reduce hallucinations. Created RAG pipelines with LangChain and LlamaIndex; fine-tuned open-source LLMs (LLaMA family) with LoRA/QLoRA for domain alignment. Automated Azure-based infrastructure using Terraform; deployed RESTful microservices (Flask/FastAPI) and GraphQL APIs; established CI/CD pipelines; tuned Snowflake data models and ELT pipelines; built internal analytics UI and governance tooling. Implemented graph-based entity extraction and RBAC for pharma data governance.

Developed end-to-end churn prediction models using Python and Scikit-learn; engineered feature pipelines from CDRs, billing history, and support logs. Achieved ROC-AUC improvement from 0.71 to 0.84 through hyperparameter tuning and feature selection. Built customer segmentation and uplift modeling to optimize campaigns; created time-series forecasts for network demand; applied NLP to support tickets for sentiment signals; deployed batch scoring for millions of records; implemented model retraining and deployment via Azure ML; established dashboards and governance; collaborated with product, marketing, and compliance teams.