GEN AI Developer

Prudhvi Raj

Gen AI/ML Engineer

*************@*****.*** +1-470-***-**** www.linkedin.com/in/prudhvi-raj-nelapatla PROFESSIONAL SUMMARY

• Over 6 years of experience in Data Science, Machine Learning, and AI solution development with specialization in Generative AI technologies.

• Experienced in applying LLMOps and safety frameworks (PromptLayer, Guardrails.ai, TruLens) to ensure

• Expertise reliability in and building compliance and deploying of GenAI LLM-applications. powered applications using LangChain, Llama, OpenAI APIs, and transformer models for tasks like document summarization, retrieval, and customer service automation.

• Strong command of NLP techniques including tokenization, named entity recognition, sentiment analysis, and text classification using frameworks like TensorFlow, Keras, LSTM, RNN, and BERT.

• Built scalable ML pipelines with Vertex AI, Kubeflow, and Docker, ensuring smooth model training, deployment, and monitoring in production environments.

• Proficient in data processing and analysis using Python, Pandas, NumPy, PySpark, and SQL, with strong data wrangling and feature engineering skills.

• Developed predictive models using scikit-learn algorithms like XGBoost, Decision Trees, KNN, and clustering techniques for both regression and classification tasks.

• Hands-on experience in cloud platforms: AWS (S3, Lambda, Redshift, SageMaker), Azure ML Studio, and Databricks for building cloud-native AI workflows.

• Skilled in building RESTful APIs with Flask, Django, and Node.js, integrating AI models into real-time applications.

• Strong full-stack experience building GenAI-powered UIs using React, Redux, TailwindCSS, and Python-based backends (Flask, FastAPI), delivering seamless real-time LLM interactions.

• Passionate about using GenAI and LLMs to build intelligent, context-aware systems that drive automation, efficiency, and data-driven decision-making.

TECHNICAL SKILLS

Languages Python, Java, JavaScript, SQL, Scala

GenAI & ML GPT-4, Hugging Face, LangChain, BERT, RAG, TensorFlow, PyTorch, Scikit-learn, Keras Web/Full Stack React, Redux, Hooks, TailwindCSS, Flask, Django, Fast API, Node.js, HTML5, CSS3, RBAC, AngularJS, Material-UI

Cloud Platforms AWS (S3, Lambda, Redshift, SageMaker, Glue), Azure ML, GCP (Vertex AI) Data & Big Data Pandas, NumPy, PySpark, Spark, Hive, HBase, Kafka, SQL Alchemy Databases MySQL, MongoDB, Snowflake, Cassandra, Cosmos DB, Oracle, Databricks Visualization Tableau, Power BI, Matplotlib, Seaborn, Plotly MLOps &

DevOps

Vertex AI, Kubeflow, Docker, Boto3, Airflow, DataStage, Git, Jenkins PROFESSIONAL EXPERIENCE

Macy’s Inc, Atlanta, GA Jan 2022 - Present

Role: Gen AI/ ML Engineer

Responsibilities:

• Developed and fine-tuned prompts for GPT-3.5, GPT-4, and Claude using OpenAI’s Assistants API and prompt templates to automate customer service, FAQ generation, and document summarization tasks.

• Built end-to-end Retrieval-Augmented Generation (RAG) pipelines integrating LangChain, FAISS, Weaviate, and Azure Cognitive Search, boosting enterprise Q&A accuracy by 35%.

• Implemented hybrid semantic + keyword search with embedding rerankers and integrated LlamaIndex for document ingestion, parsing, and chunk-based retrieval.

• Leveraged OpenAI Function Calling and Tool Use API to enable dynamic multi-step reasoning and workflow automation from LLMs in internal systems.

• Applied Guardrails.ai to ensure LLM safety, structure control, and mitigation of hallucinations in customer- facing GenAI applications.

• Designed and deployed multi-modal pipelines that incorporated PDFs, images, and structured metadata for unified retrieval using OCR tools, Azure Form Recognizer, and LangChain agents.

• Created robust LLMOps workflows with automated model versioning, monitoring (via PromptLayer, Weights

& Biases), and evaluation pipelines using EvalLangChain, TruLens, and BLEU/ROUGE metrics.

• Performed instruction fine-tuning and LoRA-based model adaptation using Hugging Face, improving factual consistency and domain alignment in generated responses.

• Conducted continuous evaluation and A/B testing of LLM variants and prompt styles using OpenAI Eval, TruLens, and custom eval scripts for scoring response helpfulness and hallucination rates.

• Engineered synthetic training data with GPT-4 turbo for edge-case scenarios and low-resource intents, boosting model generalization in production.

• Applied prompt chaining, dynamic context window resizing, and semantic memory modules for multi-turn conversations and document chaining in long-form queries.

• Created dynamic GenAI-powered user interfaces using React, Redux, TailwindCSS, and integrated conversational UI with Next.js and React Query.

• Built backend APIs in Flask and Fast API, with real-time LLM response streaming support via WebSockets and Server-Sent Events (SSE).

• Leveraged Docker, Kubernetes, and Vertex AI Pipelines for scalable model deployment and CI/CD integration.

• Implemented role-based access controls (RBAC) and secure user authentication flows using ForgeRock, enabling identity-aware LLM usage based on user roles.

• Utilized AWS Lambda, S3, Glue, and Redshift to manage GenAI data workflows for customer interaction logs and feedback collection.

• Integrated MongoDB Atlas Vector Search and PostgreSQL pgvector for hybrid retrieval and search personalization across customer interaction history. Environment: Python, LangChain, GPT-4, Claude, LlamaIndex, OpenAI Assistants API, React, Redux, TailwindCSS, Fast API, Flask, FAISS, Weaviate, Azure Cognitive Search, Azure Form Recognizer, Azure ML, Vertex AI, Docker, MongoDB Atlas, PostgreSQL, pgvector, ForgeRock, Hugging Face, Prompt Layer, Guardrails.ai, TruLens, EvalLangChain, OpenAI Eval, Next.js, Server-Sent Events (SSE), WebSocket’s August 2021 – Dec 2021

Verizon, NJ

Role: AI Developer

Responsibilities:

• Built a scalable NLP classification pipeline combining TF-IDF + SVM and later enhanced it with BERT-based LLMs to classify public apology statements by domain (e.g., celebrity, politician, corporate).

• Prototyped lightweight RAG-like context chaining with BERT embeddings and NER output before full-stack GenAI frameworks became standardized

• Implemented zero-shot classification and experimented with few-shot LLM prompting techniques using Hugging Face’s transformer models (e.g., RoBERTa, BART) for statement classification without labeled data.

• Applied core NLP techniques—tokenization, stemming, lemmatization, sentiment scoring, and POS tagging— to evaluate rhetorical structure, sentiment intensity, and public reaction trends.

• Developed and tested prompt engineering strategies using various instruction formats (e.g., Q&A, classification, summarization) to assess LLM consistency and hallucination rate.

• Built interactive embedding space visualizations using t-SNE and UMAP on BERT embeddings to understand latent topic clusters and semantic similarities between apology contexts.

• Created custom sentiment and rhetorical scoring models using Python, pandas, and rule-based algorithms to rank statements by reliability, sincerity, and moral framing.

• Conducted LLM evaluation benchmarking using BLEU, ROUGE, and cosine similarity metrics between generated and human-written summaries.

• Leveraged Azure ML Studio and Azure Notebooks to run distributed NLP jobs, automate training pipelines, and manage compute for scalable transformer training experiments.

• Developed interactive sentiment dashboards using Plotly Dash, enabling real-time filtering, comparison, and visualization of apology tone across public figures.

• Maintained full ML workflow reproducibility using Jupiter Notebooks, structured experiment logs, and Python-based pipelines for future replicability and academic publication. Environment: Python, BERT, Hugging Face Transformers, Azure ML Studio, Azure Notebooks, LangChain, TF-IDF, scikit- learn, NLTK, spaCy, Plotly, Matplotlib, Seaborn, Pandas, Jupyter, UMAP, t-SNE, ROUGE, BLEU Infosys, India May 2020 - May 2021

Role: AI Developer

Responsibilities:

• Integrated Generative AI tools (Hugging Face Transformers, OpenAI APIs) to support demand forecasting and supply chain risk analysis, enhancing decision-making processes with LLM-generated insights.

• Designed and implemented scalable deep learning models using TensorFlow for retail demand forecasting, improving prediction accuracy by 25% and reducing stockouts by 18%.

• Conducted feature engineering and exploratory data analysis on large-scale retail datasets, contributing to over $10M in annual cost savings via optimized logistics and procurement workflows.

• Implemented early-stage LLM-based summarization for executive reporting using OpenAI + LangChain, laying groundwork for advanced GenAI adoption

• Built scalable data pipelines using AWS Glue, Redshift, S3, and Lambda, ensuring seamless ingestion and transformation of multi-source supply chain data.

• Developed reusable ML/AI models for forecasting and operational risk mitigation across nodes in the supply chain, deployed via Vertex AI Pipelines and Kubeflow.

• Applied advanced predictive analytics, time-series modeling, and classification techniques using Python

(scikit-learn, XGBoost, Pandas), enabling 15% improvement in inventory turnover and reduction in logistical inefficiencies.

• Conducted LLM prompt tuning and context chaining using LangChain and domain-specific datasets to enhance inventory policy generation and executive decision support tools.

• Created scalable data warehousing and ETL solutions with AWS Redshift and Glue, reducing analytical query latency across billions of records.

• Built RESTful microservices in Flask/Django, integrating with GenAI modules and exposing APIs for supply chain automation tasks.

• Delivered interactive dashboards and executive reports using Power BI and Tableau, powered by GenAI backends for insight generation and next-best-action recommendations.

• Programmatically managed AWS infrastructure using Boto3, automating EC2, Lambda, and Redshift resources to support continuous ML deployment.

• Improved frontend and backend module performance with React (Redux, Hooks) and Python APIs, ensuring maintainability and scalability for internal analytics platforms. Environment: Python, React, Redux, Flask, Django, LangChain, OpenAI, AWS Glue, Redshift, Lambda, Spark, SQLAlchemy, TensorFlow, Keras, Hugging Face, Jenkins, Git Certifications:

• AWS Certified Developer – Associate Amazon Web Services 2023

• Certified Microsoft Azure Fundamentals (AZ-900) 2022.

• Certified Microsoft Azure AI Fundamentals (AI-900) 2022.

Contact this candidate