Home

Venkata Sai - Lead AI/ML Engineer

[email protected]

Location: Champaign, Illinois, USA

Relocation: YES

Visa: USC

Resume file: Venkata Sai_1779372236687.docx Please check the file(s) for viruses. Files are checked manually and then made available for download.

Venkata Sai Mob: (872)-216-2762. Email: [email protected] LinkedIn: www.linkedin.com/in/venkatasaig0401 EXPERIENCE: Lead AI & Data Architect with 12+ years of experience designing and delivering enterprise-scale AI, machine learning, and cloud-native data platforms across healthcare and financial services domains. Specialized in building compliant, production ready Generative AI and Agentic AI solutions using advanced RAG pipelines, knowledge graphs, multi-agent orchestration frameworks, and LLM-powered applications for clinical decision support, medical summarization, patient triage, and intelligent enterprise search. Extensive hands-on experience with Python, LangChain, LangGraph, CrewAI, Hugging Face Transformers, OpenAI APIs, vector databases, and FastAPI microservices for deploying scalable AI systems across AWS, Azure, and Kubernetes environments. Proven track record of developing AI prototypes, PoCs, and enterprise-grade solutions leveraging GPT-4, Claude, LLaMA,and fine-tuned transformer models using LoRA and QLoRA techniques. Strong expertise in LLMOps, prompt engineering, tool/function calling, AI guardrails, LangSmith, DeepEval, and MLOps practices to ensure reliable, secure, and HIPAA compliant AI deployments. Experienced in building scalable streaming and batch data pipelines, integrating FHIR/HER healthcare systems, optimizing AI inference performance, and translating complex AI capabilities into measurable business outcomes while mentoring engineering teams and guiding enterprise AI strategy and technology roadmaps. TECHNICAL SKILLS LANGUAGES: Python (3.6 3.11), Java (6 17), Scala (2.11 2.13), Kotlin, C++, Go, SQL (T-SQL, PL/pgSQL), JavaScript (ES5 ES2023), TypeScript, Node.js (12 20), Bash/Shell AI / MACHINE LEARNING / DEEP LEARNING: TensorFlow (1.15 2.14), PyTorch (0.x 2.x), Keras (2.x), Scikit-learn (0.18 1.4), XGBoost, LightGBM, Hugging Face Transformers (BERT, GPT, LLaMA, DeepSeek, Claude, Gemini), LLMs. BIG DATA & STREAMING: Apache Spark (1.6 3.5), PySpark, Spark Streaming, Spark MLlib, Hadoop (HDFS, Hive), Kafka (0.8 3.x), AWS Kinesis, GCP Dataflow, Snowflake, BigQuery ML, Vertica SQL, Databricks (Unity Catalog, MosaicAI) Generative AI & Agentic AI RAG Pipelines, LangChain, CrewAI, AutoGen, Multi-Agent Systems, LLM Fine-Tuning, Prompt Engineering, Vector Databases, Knowledge Graphs, Tool Calling, AI Guardrails, LLM Evaluation, LangSmith, DeepEval, MCP, FastAPI AI Services RISK & FINANCIAL MODELING: PD, LGD, CECL, CCAR, FRTB, SR 11-7, Basel III/IV, OCC compliance, Model Governance, Audit-ready ML, Healthcare Prior Authorization, Fraud Detection, Credit Scoring CLOUD PLATFORM: AWS: SageMaker, JumpStart, Bedrock, EC2, S3, Lambda, Step Functions, EventBridge, RDS, DynamoDB, CloudWatch Azure: Azure ML, Azure Data Factory, Azure Databricks, AKS, Log Analytics, Azure Monitor GCP: Vertex AI, BigQuery, BigQuery ML, Cloud Composer Hybrid + On-Prem Kubernetes (EKS/GKE/AKS) MODEL EXPLAIN ABILITY, API, BACKEND & GOVERNANCE: SHAP, LIME, MLflow, Model Risk Management (MRG), FastAPI, Flask, Spring (3.x Boot), Node.js/Express, REST APIs, GraphQL APIs, Microservices (Docker + K8s), AsyncIO, Event-driven architecture OBSERVABILITY, LOGGING & MONITORING: Prometheus, Grafana, CloudWatch, Splunk, AppDynamics, Azure Monitor, AIOps Monitoring Agents, Logging Pipelines, Metrics Dashboards MLOPS, DEVOPS & AUTOMATION: Docker (18 23.x), Kubernetes (EKS/AKS/GKE), Terraform, KServe, Kubeflow, Argo, Airflow (2.x), Harness, Jenkins (1.x 2.x), GitHub Actions, MLflow (1.x 2.x), CI/CD pipelines, Canary/Blue-Green Deployment, A/B Testing, Automated Retraining, Model Drift Monitoring (PSI/KS/AUC) DATABASES, VECTOR DATABASES & RETRIEVAL: PostgreSQL, MySQL, SQL Server, Oracle, MongoDB, Cassandra, Redis, DynamoDB, Vertica, Snowflake, BigQuery, PostGIS, FAISS, Pinecone, Weaviate, Qdrant, ChromaDB, Elasticsearch, ANN Search, RAG Architectures DATA VISUALIZATION & FRONTEND: Power BI, Tableau, React (15 18), Next.js (13), Streamlit, Plotly, D3.js OTHER TOOLS: ArcGIS, QGIS, GeoPandas, PostGIS, Spatial Joins, Raster/Vector Analytics, Geospatial Forecasting, Disease Outbreak Modeling, Git, GitLab CI/CD, GitHub Actions, Unity Catalog, Vault, KMS, IAM, Secrets Manager, Graph Algorithms (BFS/DFS), NetworkX, Feature Stores, Data Governance, Secure ML Wrappers (HIPAA/OCC/FedRamp), MCP (Model Context Protocols), AIOps, Real-time Simulation & Stress-Testing Tools. Professional Experience Client: AbbVie Project Role: Lead AI/ML Engineer Jul 2025 to Present Roles & Responsibilities: Worked on building enterprise Generative AI and Agentic AI systems used across R&D and clinical research teams at AbbVie. Designed end-to-end RAG pipelines including document ingestion, chunking, embeddings, vector search, and response generation for clinical and regulatory documents. Built AI workflows using LangChain and LlamaIndex with tool-calling, memory, and multi-step reasoning over scientific and regulatory data. Fine-tuned LLMs such as GPT, LLaMA, Falcon, BioBERT, and Med-BERT on biomedical datasets for literature review, summarization, and knowledge extraction. Created internal AI copilots and APIs using Python and FastAPI that integrate with enterprise platforms used by clinical and research teams. Built a GenAI system to assist in drafting clinical protocols using LLMs trained on internal and external protocol datasets. Developed the pipeline that extracts and loads protocol content into Elasticsearch and supports semantic search through a RAG-based query system. Worked on Project Magellan to build a clinical knowledge graph environment connecting clinical protocols, trial data, financial data, and site information. Built a competitive intelligence platform that uses a coding agent to automatically extract and analyze financial and scientific insights. Developed multi-agent AI systems for drug discovery workflows using a custom agent SDK with async orchestration and sandboxed code execution. Created a multimodal semantic search platform using CLIP embeddings, SBERT, and vector databases such as FAISS and Pinecone. Built NLP pipelines for pharmacovigilance workflows that analyze adverse events from PSURs, clinical safety reports, and investigational brochures. Fine-tuned LLM models using LoRA and QLoRA techniques to improve adverse event classification accuracy in regulatory safety reports. Automated PSUR safety reporting by building a GenAI pipeline that retrieves, classifies, and summarizes safety information from large document collections. Integrated agent frameworks such as AutoGen and CrewAI to automate document parsing, MedDRA coding, event classification, and compliance validation. Deployed LLM services and async APIs on AWS using EKS, Docker, GitHub Actions CI/CD, ArgoCD, and ECR. Built a prompt library platform with FastAPI APIs, Streamlit UI, and NoSQL backend to manage reusable prompts across AI applications. Optimized LLM response pipelines and reduced response latency from ~7 seconds to ~800 milliseconds for interactive AI applications. Worked closely with data engineering, compliance, and research teams to ensure AI systems follow enterprise governance, HIPAA, and FDA standards. Contributed to the data platform roadmap and helped align AI, data engineering, and analytics initiatives for next-generation clinical intelligence. Environment: Microsoft Azure, SQL, Python (Pandas, NumPy, SciPy, Matplotlib, Seaborn, Scikit-learn), R (dplyr, tidyr, ggplot2, CARET), R Studio, Tableau, Microsoft Power BI, Hive, AWS SageMaker, Amazon S3, Glue, FastAPI, Snowflake, MLFlow, Git, Excel Client: Kaiser Permanente Role: Sr Principal Engineer Aug 2024 to July 2025 Roles and Responsibilities: Led the architecture of enterprise RAG ecosystems including Unstructured.io for document processing and FalkorDB for knowledge graph integration. Set up comprehensive monitoring systems with LangSmith and DeepEval to track model drift, response quality, and potential biases in clinical outputs. Built agentic systems that use GenAI to synthesize patient history and real-time data, proposing next-best actions for care management. Guided product and clinical leadership on AI capability decisions, influencing technology roadmaps and investment priorities. Designed and deployed LangChain-based agentic workflows for medical note summarization, patient triage, and automated care coordination across departments. Architected a GenAI-driven analytics pipeline on AWS that ingests chatbot conversations to extract meaningful insights for optimizing user experience. Built HIPAA-compliant prompt orchestration pipelines for clinical decision support, prescription clarification, and automated documentation generation. Integrated multiple data sources including FHIR-based EHR systems and clinical databases for context-aware RAG responses. Integrated OpenAI GPT-4 and Claude APIs into clinical workflows for basic medical text summarization and patient intake assistance. Set up FastAPI microservices to serve AI models and connected them with React-based frontend interfaces used by clinicians. Built and maintained vector embeddings using Pinecone and Milvus for semantic search across clinical documents and medical literature. Containerized and deployed AI services on AWS Bedrock and Azure, ensuring high availability and scalability. Optimized model inference using ONNX and quantization techniques to reduce latency and infrastructure costs in production. Established guardrails and safety layers for bias detection, transparency, and clinical safety across all AI implementations. Drove adoption of agentic patterns across the organization, enabling automated reasoning and complex healthcare workflows. Partnered with compliance and legal teams to ensure all AI implementations met HIPAA, FDA, and internal regulatory standards. Mentored junior and mid-level engineers on AI best practices, production deployment strategies, and responsible AI principles. Presented technical architecture and results to executive leadership, securing buy-in for platform expansion and resource allocation. Role: Sr AI/ML Engineer Roles and Responsibilities: Dec 2021 to Aug 2024 Designed and implemented scalable streaming data pipelines using Kafka, Azure Event Hubs, and Spark Streaming for real-time analytics and operational dashboards. Architected the enterprise-scale data platform supporting analytics and AI across 10M+ patient records, enabling secure integration and insights for clinical and operational teams. Applied NLP and LLM-based techniques to accelerate cancer case ascertainment from unstructured pathology and clinical text data. Used AWS Bedrock (Claude, Titan) and Comprehend Medical to pull cancer indicators from unstructured clinical notes and pathology reports. Built and maintained ETL pipelines using Python and PySpark to process member data and clinical records from multiple source systems. Assisted in setting up Azure Databricks environments and wrote Spark jobs for data transformation and aggregation tasks. Wrote SQL queries and worked with Hive for ad-hoc data analysis requested by business and clinical teams. Participated in migrating on-premise data workloads to Azure cloud, learning cloud-native data engineering best practices. Built member segmentation models in AWS SageMaker to understand coverage patterns and what drives member behavior. Built the core data infrastructure for the organization-wide personalization platform, enabling consistent member experiences across web, mobile, and contact center channels. Developed ML-powered member clustering and segmentation models to identify overlapping coverage patterns and drivers of member behavior. Created a predictive model for coverage primacy that helped the organization achieve significant cost avoidance through better member assignments. Worked closely with data scientists to put ML models into production using Spark, H2O, and MLflow on SageMaker, and used Bedrock for fine-tuning LLMs on healthcare tasks. Optimized Spark jobs and data processing workflows to handle petabyte-scale data with improved performance and reliability. Led parts of the technical migration of legacy data systems to Azure cloud, modernizing the data architecture for AI-ready foundations. Collaborated with data scientists to productionize machine learning models using Spark, H2O, and MLflow. Role: Full stack Engineer Roles and Responsibilities: March 2021 to Dec2021 Built React.js and Redux frontend components for member-facing pharmacy and contact center applications with focus on responsive design. Led enterprise LLM deployments for customer experience modernization from discovery and SOW through launch and value realization, partnering with business and IT executives. Integrated Okta for user authentication and authorization, implementing multi-factor authentication and SSO for internal tools. Designed scalable architectures for voice and chat AI that could handle millions of member interactions securely and cost-effectively. Designed scalable architectures for voice and chat AI that could handle millions of member interactions securely and cost-effectively Translated emerging conversational AI capabilities into practical adoption plans that delivered measurable business value. Built an automated quality assurance system using ChatGPT API to analyze call transcripts, achieving over 90% agreement with human reviewers. Developed the pharmacy refill application allowing members to describe medications by appearance, purpose, or name using LLM-powered understanding. Assisted in developing RESTful APIs using Node.js and FastAPI to connect frontend applications with backend services. Participated in building a pilot application that allowed members to describe medications naturally for prescription refills. Worked with Azure Speech-to-Text APIs to support voice input features for the IVR modernization project. Fixed bugs and optimized frontend performance, ensuring consistent experiences across devices and browsers. Collaborated with UX designers and product managers to implement user-friendly interfaces for complex AI features. Participated in agile ceremonies and contributed to sprint planning and estimation. Client: USSA Project Role: Sr ML/Data Scientist Engineer March 2018 to March 2021 Roles & Responsibilities: Built data pipelines in Azure Databricks to process 50M+ financial documents (loan apps, statements, regulatory filings) for NLP and downstream ML tasks. Fine-tuned transformer models including BERT and GPT-2 for financial summarization, entity extraction, and contextual understanding using PyTorch and TensorFlow. Trained hybrid NER systems combining rule-based logic with neural networks to extract account numbers, dates, and transaction amounts at 94% precision. Developed an NLP chatbot with Python, Flask, Slack API, and MongoDB using NLTK and SpaCy for intent classification and entity recognition. Integrated Azure Cognitive Services for sentiment analysis and language understanding; applied sentence embeddings and semantic similarity for better response accuracy. Built time series models (ARIMA, SARIMA, LSTM, Prophet) for demand forecasting; final Stacked LSTM delivered 85% accuracy and cut inventory costs by 18%. Ran stationarity tests (Dickey-Fuller), ACF/PACF analysis, and seasonal decomposition to prepare clean datasets for forecasting. Set up AWS SageMaker and Azure ML pipelines to automate model training, evaluation, and deployment across dev and prod environments. Used PySpark on Glue and EMR for large-scale ETL; improved pipeline efficiency by 30% through better null handling, filtering, and data type optimization. Built FastAPI microservices for real-time model inference; handled 10K+ requests per minute with sub-100ms latency using GPU optimization. Designed database schemas, REST APIs, and middleware layers to integrate data quality tools and marketing pixels into the member-facing platform. Created Snowflake data warehouse and Power BI dashboards to track sales trends, KPIs, and product performance for business stakeholders. Applied Bayesian modeling and causal inference to improve chatbot decision-making and predict user behavior patterns. Containerized applications with Docker, set up CI/CD via Jenkins and Azure DevOps, and managed container orchestration on AKS. Tracked experiments with MLflow, versioned data with DVC, and orchestrated workflows using Airflow for reproducibility. Worked with H2O.ai, scikit-learn, and MLlib for classification, regression, clustering, and hyperparameter tuning across multiple projects. Collaborated with data science and MLOps teams on model monitoring (drift detection), documentation, and compliance with SDLC standards. Environment: Azure Databricks, Python, PyTorch, TensorFlow, BERT, GPT-2, Flask, Slack API, MongoDB, NLTK, SpaCy, Azure Cognitive Services, ARIMA, SARIMA, LSTM, Prophet, AWS SageMaker, Azure ML, AWS Glue, EMR, FastAPI, Snowflake, Power BI, Docker, Jenkins, Azure DevOps, AKS, MLFlow, DVC, Airflow, H2O.ai, scikit-learn, MLlib Client: Baird Project Role: Data scientist Engineer Sep 2016 to Feb 2018 Roles & Responsibilities: Took ownership of analytics projects from start to finish, working directly with business teams to define goals, gather requirements, and ensure the final output actually solved the business problem. Cleaned and processed messy third-party spending data using Python (Pandas, NumPy) and Excel macros to get it ready for analysis and reporting. Built Hive external tables to stage raw data and set up workflows to move it into production tables, making sure the data pipeline ran smoothly. Ran exploratory data analysis using Matplotlib and ggplot2 to understand customer behavior and spot patterns before jumping into modeling. Handled missing data using Scikit-learn s imputation techniques, so the datasets were clean and reliable for training models. Built a range of machine learning models Logistic Regression, Random Forest, KNN to predict things like customer churn and credit risk, depending on what the project needed. Trained supervised learning models on labeled data to predict credit outcomes, helping the business make better-informed decisions. Experimented with ensemble methods and algorithms like SVM, Naive Bayes, and K-Centroid clusters to improve model accuracy, often using Microsoft Azure to scale things up. Used time series analysis to study demand patterns ran stationarity tests (like Dickey-Fuller), looked at ACF/PACF plots, and built forecasting models to support rate design and planning. Created clear visualizations to explain model results like ROC curves in Matplotlib and heatmaps in Seaborn so non-technical stakeholders could understand what the data was saying. I wrote a lot of SQLS and used Python/R every day to query databases, manipulate data, and run statistical tests as part of the analysis. Designed Tableau dashboards from scratch after sitting with business users to understand what they needed to track turned those conversations into actual reports they used. Put together technical documentation and analysis reports so the team and stakeholders could follow what was built and why. Worked across different environments locally with R Studio and Jupiter, and on the cloud with AWS services like SageMaker, S3, and Glue when the data size required it. Kept the entire model lifecycle organized tracked experiments, managed datasets, and made sure everything was reproducible using tools like MLFlow, DVC, and Git. Environment: Microsoft Azure, SQL, Python (Pandas, NumPy, SciPy, Matplotlib, Seaborn, Scikit-learn), R (dplyr, tidyr, ggplot2, CARET), R Studio, Tableau, Microsoft Power BI, Hive, AWS SageMaker, Amazon S3, Glue, FastAPI, Snowflake, MLFlow, Git, Excel Client: Sutherland, New York Project Role: ETL/ Data Engineer Aug 2014 Aug 2015 Roles & Responsibilities: Extract Transform and Load data from Sources System to Data Warehouse using a combination of SSIS, T-SQL, Spark SQL. Worked on migration and conversion of data using pyspark and Spark SQL for data extraction, transformation and aggregation from multiple file formats for analyzing and transforming using Python. Ability to apply Data Frame API to complete Data Manipulation within a spark session. Created Data Quality Scripts to compare data built from spark data frame API. Design and develop ETL Integration patterns using python on spark. Analyze SQL scripts and design them by using pySpark SQL for faster performance. Performed ETL Transformation activities in SSIS and built several packages and loaded data to Data warehouse Involved in writing stored procedures in T-SQL do the transformations of the data Involved in Data Modeling Engage with business users to gather requirements, design visualizations and provide training to use self-service BI tools Environment: Python, PySpark, Apache Spark, SQL, MySQL, Informatica PowerCenter, Tableau, Azure Data Factory, Snowflake, Power BI ETL, DB2, Oracle, SSIS, Apache Hive, Apache Pig, HBase, Apache Spark, Zookeeper, Flume, Kafka, Sqoop, HDFS, Client: Cyient-Visakhapatnam, Andhra Pradesh, India Project Role: Data Analyst/ Data Engineer Aug 2012 - Jan 2013 Roles & Responsibilities: Utilized Hive and Pig for dataset transformations, deployed MapReduce jobs in Hadoop, improving processing speed by 25%, and automated infrastructure with Terraform and Ansible, reducing configuration errors by 20%. Managed complex ETL workflows with Apache Airflow, reducing failure rates by 15% and ensuring smooth pipeline operation. Designed data warehousing solutions on AWS Redshift, optimized storage and query performance, and used Hadoop and Spark for processing and Kafka for real-time analytics. Optimized Power Query using query folding and efficient loading strategies, resulting in faster data refreshing and processing times. Utilized Excel for data cleaning and transformation, leveraging pivot tables, VLOOKUP, and macros to prepare datasets for deeper analysis and pipeline integration. Designed simple and complex mappings using Informatica PowerCenter to load the data from various heterogeneous sources using different transformations like Source Qualifier, Lookup (connected and unconnected), Expression, Aggregate, Update Strategy, Sequence Generator, Joiner, Filter, Update Strategy, Normalizer, SQL, Rank and Router transformations. Environment: Python, PySpark, ETL, Hive, HDFS, Terraform, Ansible, PowerCenter, Apache Spark, SQL, Apache Pig, HBase, Apache Spark, Flume, Kafka, Sqoop, AWS Glue, AWS Redshift, HDFS, Power Query. Education: Bachelor's in computer science from University of Mumbai,2012 Keywords: cplusplus continuous integration continuous deployment artificial intelligence machine learning user interface user experience javascript business intelligence sthree active directory rlang information technology golang procedural language Kansas