AI/ML Engineer with LLM/RAG Expertise

Sai Kumar Nune

New Haven, CT – ***** — ***************@*****.*** — +1-636-***-**** — LinkedIn

Professional Summary

AI/ML Engineer with 5+ years of experience developing and deploying LLM-driven systems, multi- agent RAG architectures, and semantic search applications for enterprises across healthcare, finance, and telecom. Skilled in designing LangChain and LangGraph-based pipelines that integrate GPT-4, LLaMA2, and vector databases (FAISS, ChromaDB) for context-aware automation and knowledge re- trieval. Experienced in building FastAPI-powered AI microservices, implementing MLOps pipelines with Docker, Kubernetes, and CI/CD, and fine-tuning models using Hugging Face Transformers and Tensor- Flow. Adept at translating business goals into scalable AI workflows, reducing latency, and improving accuracy through prompt optimization, metadata retrieval, and responsible AI practices. Category Tools & Technologies

AI/ML & NLP GPT-3/4, LLaMA2, T5, BERT, SBERT, LangChain, LangGraph, Hugging Face, RAG, Prompt Engineering, TensorFlow, PyTorch, Scikit-learn, XGBoost, SpaCy, NLTK, Semantic Search, NER, Summarization, Tool-using Agents, Transformers, ANN, CNN, Ensemble Methods, Recommender Systems, Few-shot Learning Data & Engineer-

ing

Python (Pandas, NumPy, Matplotlib), SQL, NoSQL, MySQL, PostgreSQL, Mon- goDB, Airflow, REST APIs, FAISS, ChromaDB, Neo4j, Pinecone, Spark, Flask Cloud & DevOps AWS (SageMaker, Lambda, EC2, S3), Azure, Docker, Kubernetes, OpenShift, Git/GitHub, GitHub Actions, Bitbucket, Harness, CI/CD Visualization Power BI, Tableau, VS Code, Excel

Computer Vision OpenCV, Image Classification, Object Detection, Vision Transformers Educational Details

Sacred Heart University, Fairfield, Connecticut Jan 2024 – Mar 2025 Master of Science (M.S.) in Computer Science – CGPA: 3.76/4 Lakireddy Bali Reddy College of Engineering, India Dec 2021 Bachelor of Technology in Electronics and Communication Engineering – CGPA: 7.59 Professional Experience:

Responsive – Remote (AI Engineer) July 2024 – Present Roles & Responsibilities:

• Designed and deployed enterprise-grade LangGraph-powered GPT-4 chatbots for internal automa- tion, improving service response time by 35% through multi-agent orchestration and intelligent workflow management.

• Built and integrated memory-aware agents and multi-tool integrations using LangChain React Agent framework, orchestrating multi-step reasoning workflows for real-time, personalized, context- driven user interactions.

• Engineered robust Retrieval-Augmented Generation (RAG) systems with LangChain, FAISS, ChromaDB, Pinecone, and SQL databases to enhance enterprise knowledge access and deliver context-aware, factually accurate responses.

• Increased ingestion throughput by 40% through token-optimized chunking pipelines for PDFs and CSVs with intelligent data preprocessing using Python, Pandas, and NumPy.

• Designed graph-based Q&A agents using Neo4j graph database, implementing semantic search and improving real-time contextual retrieval for complex query workflows across distributed knowledge bases.

• Developed and optimized prompt engineering strategies for GPT-4 and LLaMA2, enabling domain-specific automation, personalized content generation, and improved model response quality.

• Applied few-shot learning techniques and advanced NLP pipelines using Transformers, BERT, and Hugging Face libraries for NER, summarization, and classification tasks.

• Created and maintained CI/CD pipelines using Docker, Kubernetes, OpenShift, Git/GitHub, GitHub Actions, Bitbucket, and Harness for streamlined model deployment, version control, and continuous monitoring.

• Deployed and managed NLP and CV models in production environments on AWS SageMaker, Lambda, EC2, and S3 with high availability and scalability.

• Built REST APIs using Flask for model serving and integration with front-end applications, ensuring low-latency responses.

• Incorporated responsible AI practices by implementing model explainability, fairness checks, bias detection, and compliance measures in deployed systems.

• Collaborated in cross-functional Agile teams using tools like VS Code for development, ensuring align- ment with business goals and measurable impact.

Tech Stack: GPT-3/4, LLaMA2, LangChain, LangGraph, LangChain React Agent, FAISS, ChromaDB, Neo4j, Pinecone, SQL, NoSQL, MongoDB, PostgreSQL, MySQL, OpenAI API, Python, Pandas, NumPy, Matplotlib, Flask, REST APIs, Docker, Kubernetes, OpenShift, Git, GitHub, GitHub Actions, Bitbucket, Harness, AWS (SageMaker, Lambda, EC2, S3), CI/CD, RAG, Prompt Engineering, Few-shot Learning, Vector Databases, Semantic Search, Multi-Agent Orchestration, Tool-using Agents, Transformers, BERT, Hugging Face, NER, Summarization, Real-time LLM Applications. Wipro – Hyderabad, India (GenAI / ML Engineer) Dec 2021 – Mar 2024 Roles & Responsibilities:

• Designed and deployed Retrieval-Augmented Generation (RAG) systems using LangChain, FAISS, and ChromaDB, integrating LLMs (GPT-4, LLaMA2, T5) to support healthcare claims automation, medical code extraction, and intelligent document summarization.

• Reduced manual triage by 35% through optimized RAG workflows, improved metadata retrieval pipelines, and semantic search capabilities for claim fraud detection and evidence retrieval.

• Fine-tuned LLMs (GPT-4, LLaMA2, T5) for summarization and document intelligence pipelines using prompt engineering, few-shot learning, and transfer learning techniques.

• Built advanced NLP pipelines using AWS SageMaker, BERT, T5, SBERT, spaCy, and NLTK for clinical code recommendations, NER, textual entailment, and intelligent chatbot responses.

• Developed and deployed deep learning and neural network models (ANN, CNN) using Ten- sorFlow and PyTorch with hyperparameter tuning, regularization, and backpropagation to improve model accuracy.

• Trained computer vision models with PyTorch, TensorFlow, and OpenCV for document-based visual information extraction, dental image analysis, document verification, and object detection tasks.

• Applied machine learning techniques including Regression, Classification, Clustering, En- semble Methods, XGBoost, and Scikit-learn for predictive analytics, behavioral modeling, and segmentation.

• Led deployment optimization, reducing inference latency by 20% through fine-tuning deployment strate- gies on AWS Lambda, EC2, and S3 in collaboration with DevOps teams.

• Replicated and extended features of WiNow, Wipro’s internal AI chatbot platform—built chatbot workflows for timesheet entry, leave requests, approvals, and ticket status tracking using conversational AI, demonstrating up to 80% reduction in manual effort and 50% faster turnaround.

• Engineered and optimized ETL workflows using Apache Airflow and Spark to ingest, transform, and process large-scale structured and unstructured datasets from healthcare repositories and SQL databases.

• Built and maintained CI/CD pipelines with Docker, Kubernetes, Git, and Azure cloud-native tools to automate retraining, testing, and production deployment of AI models.

• Developed automated data pipelines with Python, NumPy, Pandas, and Matplotlib for data preprocessing, feature engineering, and visualization.

• Created interactive dashboards and data visualizations using Tableau, Power BI, and Excel to communicate model insights and business metrics to stakeholders. Tech Stack: LangChain, FAISS, ChromaDB, GPT-4, LLaMA2, T5, BERT, SBERT, spaCy, NLTK, Ten- sorFlow, PyTorch, Scikit-learn, XGBoost, OpenCV, AWS (SageMaker, Lambda, EC2, S3), Azure, Docker, Kubernetes, Git, Apache Airflow, Spark, Python, Pandas, NumPy, Matplotlib, SQL, MySQL, PostgreSQL, Tableau, Power BI, Excel, Prompt Engineering, Few-shot Learning, NER, Summarization, ANN, CNN, En- semble Methods, Regression, Classification, Clustering, Object Detection, Image Classification, Vision Trans- formers, Semantic Search, RAG.

Zensar Technologies – Hyderabad, India (Associate AI Engineer) Dec 2020 – Nov 2021 Roles & Responsibilities:

• Created classification, regression, and anomaly detection models using ensemble methods, XGBoost, and Scikit-learn with retraining workflows using Apache Spark and Airflow.

• Deployed transformer-based NER and classification models built with TensorFlow, PyTorch, and Hugging Face Transformers on AWS and Azure cloud platforms for finance and telecom domains.

• Developed automated data pipelines and ETL workflows with Apache Airflow, Python, NumPy, and Pandas to process large-scale structured and unstructured datasets from SQL, NoSQL, MySQL, PostgreSQL, and MongoDB databases.

• Applied NLP techniques using spaCy, NLTK, and BERT for text analytics, NER, sentiment analysis, and summarization tasks.

• Built executive dashboards and interactive data visualizations using Power BI, Tableau, and Excel to visualize model performance and business metrics, communicating insights to stakeholders.

• Implemented CI/CD pipelines using Git/GitHub for version control and automated model deploy- ment.

• Utilized VS Code for development and collaborated with cross-functional teams to ensure model ex- plainability, alignment with business goals, and measurable impact of AI solutions.

• Designed and built recommender systems using collaborative filtering and content-based approaches for personalized user experiences.

Tech Stack: Python, TensorFlow, PyTorch, Scikit-learn, XGBoost, Hugging Face, Transformers, Apache Spark, Airflow, NumPy, Pandas, Matplotlib, spaCy, NLTK, BERT, AWS, Azure, Git, GitHub, Power BI, Tableau, Excel, VS Code, SQL, NoSQL, MySQL, PostgreSQL, MongoDB, Regression, Classification, Cluster- ing, Ensemble Methods, NER, Summarization, Recommender Systems. Key Professional Projects

Healthcare RAG Bot:

Built GPT-4 LangChain Q&A system with FAISS vector database for healthcare triage automation, implementing semantic search and prompt engineering, cutting response time by 40%. Multi-Agent RAG System for Claims Processing:

Engineered LangGraph-based orchestration system integrating GPT-4 and LLaMA2 with FAISS and ChromaDB for automated healthcare claims fraud detection. Reduced manual review time by 35% and improved detection accuracy by 22% through optimized RAG workflows and metadata retrieval pipelines.

Auto-Form Extraction:

Created Vision Transformer-based document extractor using PyTorch and OpenCV, improving accuracy from 79% to 92% through deep learning techniques.

Chat Agent Orchestration:

Built LangGraph-based multi-agent system with tool-using agents and memory-aware capabilities, reducing SLA violations by 30%.

Awards and Certifications:

• NVIDIA-Certified Associate: Generative AI LLMs

• DeepLearning.AI – Generative AI with LLMs (Coursera) – 2025

• Udemy – Artificial Intelligence A-Z – 2025

• Udemy – Certified Machine Learning Engineer – 2024

Contact this candidate