Santosh Kumar Kuricheti

Projects and Contributions

AI SDK and CLI Framework for AWS Bedrock

• Designed and developed a Python-based SDK and CLI framework (LangChain-like) to simplify the development and deployment of LLM-powered applications on AWS Bedrock.
• Built modular providers for LLMs, embeddings, OpenSearch vector store, S3 object storage, Guardrails, logging, and Langfuse-based observability.
• Implemented a suite of CLI commands (init, run, build, test, deploy, health, monitor) to support project scaffolding, local development, containerization with Podman/Docker, image management via JFrog Artifactory, and deployment to Kubernetes with automated endpoint provisioning.
• Standardized the AI application lifecycle from prototyping to production, ensuring secure, observable, and enterprise-ready deployments.

Advanced RAG Pipeline (PMI Project)

• Developed an enterprise-grade RAG pipeline with document boosting, reranking, and query enrichment.
• Optimized for regulatory and compliance search in financial document collections.
• Improved retrieval accuracy and response reliability for business-critical queries.

Multimodal RAG using Gemini Vision Models

• Built a multimodal RAG system that combined text and image inputs using Gemini Vision models.
• Enabled processing of insurance claims, scanned contracts, receipts, and reports with high accuracy.
• Improved document understanding by integrating image interpretation with textual retrieval.

Financial Document RAG for Compliance QA

• Designed a financial compliance assistant for regulatory audits and QA.
• Indexed regulatory filings, policies, and financial reports in a vector store for efficient search.
• Delivered a question-answering solution tailored for compliance officers and auditors.

Agentic Architectures with LangGraph and Google ADK

• Built agent-based architectures capable of dynamic reasoning and tool orchestration.
• Integrated SQL agents, document retrievers, and status checkers for enterprise workflows.
• Designed multi-turn customer support bots with contextual reasoning and decision-making.

LLM + OCR Pipelines for Document Intelligence

• Automated the ingestion of insurance claims, invoices, and bills into structured SQL tables.
• Used OCR tools (PyMuPDF, pdfplumber, Tesseract) to handle scanned and mobile-captured documents.
• Combined Gemini Vision with RAG pipelines for hybrid document and text intelligence.

Natural Language to SQL Agents

• Built NL-to-SQL translation agents using LangChain, Pandas, Gemini, and LangGraph.
• Enabled business users to query datasets in plain English.
• Delivered interactive, dashboard-like responses with summaries, charts, and insights.

Automatic Ticketing Tool

• Automated the extraction and classification of support tickets from Excel files.
• Used LangChain + OpenAI for ticket categorization and prioritization.
• Improved workflow efficiency for IT and customer service teams.

Customer Support Bot with Agentic RAG

• Developed an end-to-end agentic support bot for enterprise customer service.
• Combined multi-turn reasoning, contextual retrieval, and tool usage.
• Reduced manual support overhead by automating ticket handling and responses.

Document Intelligence and Recommendation System

• Created a unified document intelligence platform combining OCR, RAG, and classification pipelines.
• Enabled seamless processing of both text and image documents.
• Added a recommendation engine for related documents, cases, and next-best actions.

Document Classification and Data Grouping Pipelines

• Built LLM-driven classification workflows for document tagging, ticket categorization, and data grouping.
• Applied a mix of prompt engineering and fine-tuning to achieve high accuracy on noisy data.

Enterprise Deployment and Interfaces

• Developed Streamlit-based interfaces for business demos, prototyping, and validation.
• Deployed solutions on AWS (S3, Lambda) and Azure AI Foundry for scalability.
• Managed secure access, storage, and event triggers for both batch and live data pipelines.

• Participated in meetings to discuss supply chain improvements and workflow efficiencies.
• Assisted in analyzing inventory levels to ensure optimal stock availability.
• Conducted research on best practices in supply chain management for process improvement.
• Implemented strategies to reduce costs in the supply chain system by streamlining processes.