N G Bhanupriya

Lead AI Engineer

Health.gov • Temps plein

May 2025 - Present • 1 yr 2 mos

1. Conversational Chatbot Assistant (LangGraph + React): Interactive multi-turn conversational AI system Architected a stateful conversational agent using LangGraph, modeling dialogue flow as a graph to handle branching logic, tool routing, and multistep reasoning across turns. Built a responsive React-based frontend with streaming responses, integrated with FastAPI backend endpoints for real-time interaction. Implemented persistent conversation state and context management across sessions to enable coherent, long-running interactions. Integrated pgvector as the vector database for storing embeddings and powering retrieval-augmented generation (RAG), enabling semantic search over the knowledge base. Designed conditional routing and node-based workflows in LangGraph for dynamic decision-making based on user intent. Integrated tool-calling capabilities to extend the chatbot with external actions and data retrieval. Deployed the system on Red Hat OpenShift for container orchestration, scalability, and enterprise-grade API access. 2. Resume Standardization System (Google ADK, Multi-Agent Orchestration): Automated resume reformatting to corporate template Designed a multi-agent orchestration pipeline using Google's Agent Development Kit (ADK) that ingests resumes in arbitrary formats and outputs a standardized CitiusTech-template version. Decomposed the workflow across specialized agents (parsing/extraction, section classification, formatting, and QA validation) coordinated through orchestrated agent handoffs. Implemented structured data extraction to capture experience, skills, education, and accomplishments from unstructured resume inputs. Automated template-compliant formatting to ensure consistent, standardized output across all processed resumes. Reduced manual resume-formatting effort and turnaround time while ensuring template consistency at scale. Skills Used: LangGraph, LangChain, Google ADK, Multi-Agent Orchestration, React, Azure OpenAI, RAG, Prompt Engineer

Data Scientist

Caviar, • Temps plein

May 2024 - May 2025 • 1 yr

Designed and deployed GenAI-powered conversational agents using LangChain and Azure OpenAI, enabling contextual Q&A and document summarization for internal teams. Integrated LangChain agents with vector databases (FAISS, Azure Cognitive Search) to enable retrieval-augmented generation (RAG) from PDF and knowledge base inputs. Built a multi-tool AI assistant using LangChain tools (calculator, search, SQL agent) for handling internal dashboard queries and forecasting metrics. Created prompt pipelines for summarization, extraction, and content generation workflows, reducing manual analysis time by 40%. Used MLflow and Azure ML to version and monitor LLM pipelines, ensuring reproducibility and stability of production GenAI workflows. Collaborated with frontend developers to embed LangChain-based chat widgets into internal dashboards and customer support tools. Worked with LangGraph for agentic workflows where multiple LLM chains managed dynamic decision-making across structured and unstructured sources. Documented and deployed LangChain pipelines via FastAPI endpoints for scalable integration into existing microservices Skills Used: LangChain, LangGraph, Azure OpenAI, Prompt Engineering, FAISS, Python, MLflow, FastAPI, Azure ML, NLP

Data Scientist

Recruit Smarter • Temps plein

Jan 2018 - Apr 2024 • 6 yrs 3 mos

Developed a supervised ML model to evaluate and predict job quality perception based on human capital metrics, enabling data-driven recruitment strategies and workforce planning. Performed extensive data cleaning, preprocessing, and imputation on raw customer datasets using Pandas and ML-based missing value handling, improving data reliability by ~30%. Built a Linear Regression model to analyze demand-to-fulfillment ratios, enabling the team to anticipate demand gaps and reduce client-side attrition. The model improved resource allocation efficiency by 15%. Designed a Logistic Regression model to forecast employee deployment risk based on recruitment inefficiencies and early attrition patterns—helping stakeholders reduce early dropouts and optimize workforce contracts. Engaged directly with clients for model iterations, feedback integration, and deployment, fostering stronger alignment between business needs and ML outputs. Skills Used: Python, Pandas, NumPy, Scikit-learn, Tableau, ML Models (Logistic & Linear Regression), Power BI, Tableau, Python (for preprocessing), Excel