Generative AI & LLM Systems Engineer

BHAVANA UPPALAPATI

AI Engineer Generative AI LLM Systems

Kansas City, MO +1-816-***-**** *******.****@*****.*** https://www.linkedin.com/in/bhavanaa19/

Professional Summary:

AI Engineer specializing in Generative AI, LLM systems, and multi-agent architectures, with 4+ years of experience building and deploying production-grade AI applications. Experienced in designing end-to-end RAG pipelines and LangGraph-based agentic workflows, leveraging tools such as LangChain, FAISS, and OpenSearch for intelligent retrieval and reasoning. Proficient in developing scalable AI systems using FastAPI, Docker, and AWS (ECS, Lambda, S3), with strong expertise in LLMOps, structured prompting, and evaluation using the RAGAS framework. Experienced working with LLMs including OpenAI GPT-4 and Claude-style models for reasoning, retrieval, and agent-driven workflows. Additionally skilled in machine learning, NLP, and distributed data processing using PySpark, with a proven ability to deliver robust, production-ready AI solutions.

Tools & Technologies

Languages: Python, C, C++, SQL, TypeScript

Generative AI & LLM: LangChain, LangGraph, LangSmith, GPT-4o, GPT-3, OpenAI API, Hugging Face Transformers, Sentence Transformers, FAISS, Chroma, RAGAS

Machine Learning & Deep Learning: Scikit-learn, TensorFlow, PyTorch, XGBoost, LightGBM, CatBoost, Random Forest, Logistic Regression, SVM, Naive Bayes, KNN, Optuna, SHAP, LIME, MLflow

NLP & Computer Vision: spaCy, NLTK, TF-IDF, Word2Vec, Bag of Words, OpenCV, CNNs, Azure LUIS, Azure Bot Framework, Azure Cognitive Services

Data Engineering & Databases: PySpark, MLlib, Pandas, NumPy, PostgreSQL, MySQL, MongoDB, DynamoDB, ETL Pipelines, AWS Step Functions

APIs & Backend: FastAPI, Flask, SQLAlchemy, OAuth2, RESTful APIs, Celery, Redis, Streamlit, Pytest

Cloud & DevOps: AWS (SageMaker, ECS, EC2, Lambda, S3, Textract, CloudWatch), Azure (ACI, Cognitive Services, Bot Framework), Docker, GitHub Actions, Jenkins, Git, CI/CD

Visualization & BI: Tableau, Plotly, Matplotlib, Seaborn

Professional Experience:

Epsilon Dec 2025 – Current

AI Engineer Chicago, IL

Delivered end-to-end AI solutions from ideation to production, collaborating with stakeholders to identify high-impact use cases and translate business requirements into scalable AI systems.

Developed intent classification layers using large language models to detect and categorize user request types at runtime, dynamically dispatching each request to the appropriate specialized agent or tool, improving resolution accuracy across diverse query patterns.

Constructed end-to-end RAG pipelines using LangChain, integrating FAISS for local vector similarity search and OpenSearch for distributed semantic indexing, applying Sentence Transformers for embedding generation with citation tracking, achieving ~90% answer accuracy (RAGAS-based evaluation).

Applied semantic reranking using cross-encoder models on FAISS retrieval results to rescore and reorder candidate chunks by contextual relevance before passing them to LLMs, improving top-k precision and overall answer quality.

Designed tool-calling agent workflows where LLM-powered agents dynamically invoke tools such as SQL queries via SQLAlchemy, document retrieval via OpenSearch, and external REST APIs within a unified reasoning loop managed by LangGraph.

Built document extraction pipelines using AWS Textract to process PDFs and scanned images, transforming structured outputs into OpenSearch indexes for downstream RAG and semantic search workflows.

Developed stateful conversational agents using LangChain (ConversationBufferMemory) to maintain multi-turn context, delivering real-time token-streaming responses through FastAPI to React and Streamlit frontends.

Designed structured prompting frameworks and implemented LLMOps observability using LangSmith and MLflow to track prompt performance, latency, and token usage

Evaluated RAG pipeline performance using the RAGAS framework, measuring faithfulness, context recall, and answer relevancy to identify retrieval and generation gaps, achieving ~80–88% scores across evaluation metrics and iteratively refining system performance.

Designed horizontally scalable APIs using ECS and load-balanced microservices. Optimized distributed task execution using Celery workers and queue partitioning for high-throughput processing.

Built and deployed FastAPI-based LLM inference APIs on AWS ECS using Docker, implementing Redis caching and Celery-based asynchronous processing to support high-concurrency workloads handling thousands of requests.

Established CI/CD pipelines with GitHub Actions and optimized system performance through caching, prompt tuning, and retrieval improvements, reducing latency and operational costs.

Project Client: Maybank Aug 2022 – Jul 2024

Senior Software Engineer Hyderabad, IN

Developed and deployed RAG (Retrieval-Augmented Generation) systems using LangChain and OpenAI GPT-3 for intelligent document retrieval and question-answering, achieving 85–90% answer accuracy, and implemented semantic search using Chroma and FAISS vector databases for efficient similarity matching across large document collections.

Built supervised learning models for classification and regression using tree-based algorithms and deep learning frameworks (TensorFlow, PyTorch) for forecasting and pattern recognition use cases.

Built end-to-end data pipelines using Python and SQL, processing large-scale datasets with PySpark on distributed systems, and worked with PostgreSQL, MySQL, and MongoDB for data storage and retrieval.

Developed NLP solutions using Hugging Face Transformers for fine-tuning pre-trained language models, implemented text processing pipelines using spaCy and NLTK, integrated Azure LUIS for intent recognition, and built conversational interfaces using Azure Bot Framework.

Performed hyperparameter tuning using Optuna (Bayesian optimization) and Grid Search with Randomized Search, and applied SHAP and LIME for model interpretability and explaining predictions to stakeholders.

Deployed machine learning models using AWS SageMaker, containerized applications using Docker, deployed on Azure ACI for scalable inference, and built RESTful APIs using FastAPI and Flask to serve model predictions with average inference latency under ~300 ms.

Implemented computer vision solutions using OpenCV, built CNN architectures with TensorFlow, and integrated Azure Cognitive Services APIs for image analysis, OCR text extraction, and content moderation.

Performed exploratory data analysis using Pandas and NumPy, created visualizations using Matplotlib, Seaborn, and Plotly, and developed business intelligence reports using Tableau for non-technical stakeholders.

Automated machine learning workflows using MLflow, set up CI/CD pipelines with GitHub Actions, and orchestrated workflows using AWS Step Functions and Docker to enable scalable and reliable model deployment.

Developed predictive models using tree-based algorithms and collaborated with engineering teams to integrate models into production systems with clean code and version control practices

McKesson Nov 2021 – Jul 2022

Software Engineer Hyderabad, IN

Built and maintained high-performance applications using Python and FastAPI to develop RESTful APIs for secure and scalable data access, implementing OAuth2 authentication, role-based access control, and structured logging, improving API response efficiency by 25%.

Designed and developed ETL pipelines using Pandas, NumPy, and PySpark to process and transform large-scale structured and unstructured datasets, enabling scalable data wrangling, cleaning, profiling, and aggregation for downstream analytics, and integrated data into PostgreSQL, optimizing schema design and indexing to reduce query latency by 20–35%.

Developed and evaluated machine learning models using Scikit-learn and TensorFlow, supporting end-to-end workflows from data preprocessing to model training and evaluation, and performed NLP-based feature engineering for text data using vectorization techniques.

Deployed and managed machine learning models using AWS SageMaker, EC2, and Lambda, leveraging S3 and DynamoDB for scalable data storage and model artifact management, and implemented asynchronous and distributed data processing using Celery and PySpark MLlib to improve pipeline performance and reduce processing latency.

Developed data visualization dashboards using Matplotlib, Seaborn, and Tableau, enabling stakeholders to derive actionable insights and improve decision-making efficiency by 25%.

Built and maintained CI/CD pipelines using Jenkins and Git, applying Test-Driven Development (TDD) with Pytest, and containerized applications using Docker, ensuring deployment consistency, scalability, and reliable delivery within an Agile (Scrum) environment.

Project

Intelligent Incident Analysis & Resolution Assistant

Designed and developed an AI-driven system to analyze IT incident logs and support tickets, enabling automated root cause identification and resolution recommendations.

Built a context-aware retrieval pipeline to match incoming incidents with historical cases, improving troubleshooting accuracy and reducing manual debugging effort.

Implemented semantic search and reasoning workflows using embedding-based similarity and LLM-based analysis to generate actionable insights for issue resolution.

Educational Details:

University of Central Missouri – Kansas City, MO

Master of Science in Computer Science

Contact this candidate