Senior GenAI & ML Engineer

Tirumala Teja

Senior Data & Gen AI/ML Engineer

*********@*****.***

+1-940-***-**** LinkedIn

Python SQL PySpark Apache Spark Scikit-learn XGBoost PyTorch TensorFlow Hugging Face LLMs RAG GraphRAG LangChain LangGraph AI Agents MCP A2A FastAPI MLflow Docker Kubernetes Terraform AWS SageMaker Azure OpenAI Azure ML GCP Vertex AI Databricks Snowflake BigQuery Neo4j Redis FAISS ONNX Runtime NVIDIA Triton LangSmith

PROFESSIONAL SUMMARY

Data-driven AI/ML and Generative AI Engineer with 10+ years of experience progressing from full-stack Python and data engineering into data science, machine learning, deep learning, MLOps, and enterprise GenAI delivery across production environments.

Strong hands-on experience with Generative AI, Agentic AI, RAG, GraphRAG, NLP, and document intelligence, including multi-agent workflows, tool calling, MCP, Agent to Agent(A2A) Protocol integrations, retrieval optimization, and governed enterprise AI applications.

Experienced in building machine learning and deep-learning solutions for classification, regression, forecasting, anomaly detection, customer analytics, NLP, computer vision, entity extraction, and predictive modeling.

Built end-to-end ML pipelines covering data ingestion, preprocessing, feature engineering, model training, tuning, validation, deployment, scoring, monitoring, and retraining across batch and real-time workloads.

Strong background in large-scale data engineering and analytics using Python, SQL, PySpark, Spark, BigQuery, Databricks, Snowflake, S3, ADLS, ETL/ELT pipelines, and streaming data workflows.

Extensive multi-cloud experience across AWS, Azure, and GCP, using services such as SageMaker, EKS, Glue, EMR, Azure ML, Azure OpenAI, Databricks, AKS, Vertex AI, BigQuery, Dataflow, and Dataproc.

Deep experience with MLOps and LLMOps, including MLflow, SageMaker Pipelines, Azure ML, Docker, Kubernetes, Terraform, CI/CD, model registries, automated validation, rollback, drift monitoring, and production observability.

Built scalable AI/ML APIs and microservices using FastAPI, Flask, Django, REST APIs, and Java/Go-integrated services for model inference and enterprise application integration.

Hands-on with relational, NoSQL, graph, warehouse, cache, and vector-search platforms, including PostgreSQL, MySQL, Aurora PostgreSQL, MongoDB, Redis, BigQuery, Neo4j, Azure AI Search, and FAISS.

Built and optimized GPU-accelerated AI workloads using NVIDIA GPUs, CUDA/cuDNN, mixed-precision training, Triton Inference Server, ONNX Runtime, model quantization, batching, and scalable CPU/GPU inference patterns.

Applied model evaluation, explainability, governance, and responsible AI practices using SHAP/LIME, RAGAS, LangSmith, drift monitoring, data-quality checks, model-risk controls, and PII/PHI protection.

Proven ability to work across data science, software engineering, platform, product, security, compliance, and business teams, translating requirements into scalable AI/ML solutions while supporting Agile delivery, testing, troubleshooting, documentation, and production operations.

TECHNICAL SKILLS

Programming & Scripting

Python, SQL, PySpark, JavaScript, TypeScript, Java, Go, Shell/Bash, Linux

AI/ML & Statistics

Scikit-learn, XGBoost, LightGBM, Random Forest, Logistic Regression, K-Means, Pandas, NumPy, SHAP, LIME, A/B Testing, Hypothesis Testing, Feature Engineering, Time-Series Forecasting, Anomaly Detection

Generative AI & LLMs

LLMs, RAG, GraphRAG, Agentic AI, LangChain, LangGraph, CrewAI, Prompt Engineering, MCP, A2A, PEFT, LoRA, QLoRA, RAGAS, NeMo Guardrails, LLM-as-a-Judge, Prompt Engineering, RLHF

Deep Learning & Computer Vision

PyTorch, TensorFlow, Hugging Face Transformers, CNNs, Vision Transformers (ViT), OpenCV, BioBERT, ClinicalBERT, FinBERT, NVIDIA CUDA/cuDNN, Mixed-Precision Training, Optical Character Recognition (OCR).

Cloud Platforms

AWS: (SageMaker, Bedrock, S3, EC2, EKS, ECR, Glue, EMR, Lambda, Kinesis, MSK, Redshift, Athena, CloudWatch)

Azure: Azure OpenAI, Azure ML, AKS, ADLS Gen2, Azure SQL, Azure AI Search, Azure Functions, API Management, Azure DevOps

GCP: Vertex AI, Gemini, BigQuery, Dataflow, Dataproc, GKE, Cloud Storage, Document AI, AI Platform, cloud composer

Big Data & Data Engineering

Apache Spark, Kafka, PySpark, Azure Databricks, Snowflake, ETL/ELT Pipelines, Feature Pipelines, Batch & Streaming Processing, Apache Airflow

Databases, Warehousing & Vector Search

PostgreSQL, MySQL, Amazon Aurora PostgreSQL, DynamoDB, MongoDB, Redis, Snowflake, Azure SQL, BigQuery, Neo4j, Azure AI Search, FAISS

Backend & API Development

FastAPI, Flask, Django, Django REST Framework, Pydantic, REST APIs, Microservices, Java Services, Go Services, Azure Functions, Azure API Management, Amazon API Gateway

Frontend & Visualization

React, TypeScript, JavaScript, HTML5, CSS3, Bootstrap, Streamlit, Power BI, Tableau

Tools & Methodologies

Jira, Confluence, Git, GitHub, Agile/Scrum, Linux, Shell Scripting, Unit Testing, Code Reviews

MLOps, LLMOps & DevOps

MLflow, SageMaker Pipelines, Azure ML, Docker, Kubernetes (EKS/AKS/GKE), NVIDIA Triton Inference Server, Terraform, Jenkins, Azure DevOps, Git, GitHub, CI/CD, GitHub Actions, DVC

Model Optimization & Inference

vLLM, NVIDIA Triton, ONNX Runtime, GPU Acceleration, Model Quantization, CUDA/cuDNN, Mixed Precision, Batching, Autoscaling, Hyperparameter Tuning

Monitoring, Evaluation & Governance

SageMaker Model Monitor, CloudWatch, Prometheus, Grafana, LangSmith, RAGAS, SHAP, LIME, Model/Data Drift, Data Quality, Model Validation, PII/PHI Controls

PROFESSIONAL EXPERIENCE

Client: BNY Mellon – New York, NY Jan 2025 - Present

Role: Senior Data & Generative AI/ML Engineer

Overview:

Worked on enterprise AI/ML initiatives spanning agentic AI, RAG/GraphRAG, document intelligence, NLP, model fine-tuning, anomaly detection, and MLOps across Azure, GCP, and AWS. Delivered financial AI solutions with a focus on secure retrieval, model governance, AML analytics, PII/MNPI controls, auditability, and reliable production deployment.

Key Responsibilities:

Designed an enterprise multi-agent RAG platform using Azure OpenAI, LangChain, LangGraph, and Azure AI Search, delivering good context precision on curated evaluations for financial, operational, policy, and regulatory knowledge.

Built stateful agent workflows with MCP, A2A, tool calling, memory, and human-in-the-loop approvals, integrating APIs, databases, and enterprise systems through Azure Functions and API Management for controlled financial workflow automation.

Improved retrieval using hybrid search, semantic chunking, metadata filtering, reranking, Hugging Face embeddings, and FAISS-based ANN evaluation, with GraphRAG and Neo4j modeling relationships across financial entities.

Developed production LLM inference services using Python, FastAPI, Pydantic, vLLM, and lightweight Go services, with Redis caching, quantization, and GPU optimization to reduce p95 inference latency by ~30% on AKS.

Built React/TypeScript interfaces with FastAPI services for agentic RAG workflows, integrating Azure SQL, Neo4j, and Azure AI Search through Azure API Management for streamlined analyst access and workflow execution.

Implemented LLMOps, LLM evaluation and alignment workflows using RAGAS, LangSmith, LLM-as-a-Judge, Reinforcement Learning from Human Feedback (RLHF), and NeMo Guardrails, incorporating human feedback to improve response quality while enforcing PII/MNPI controls, prompt security, audit logging, and production guardrails.

Built a document-intelligence platform using Google Document AI, Optical Character Recognition (OCR), Vertex AI, Gemini, and Cloud Storage to extract and structure content from credit agreements, regulatory disclosures, financial statements, and operational documents.

Fine-tuned FinBERT and Hugging Face transformer models using PEFT, LoRA, and QLoRA, achieving good financial NER across counterparties, covenants, obligations, risk terms, dates, and regulatory concepts.

Developed NLP pipelines for NER, document classification, semantic similarity, summarization, and contextual extraction using FinBERT, transformer embeddings, OCR outputs, and domain-specific validation rules.

Implemented RAG and GraphRAG using Vertex AI embeddings, Vertex AI Vector Search, and Neo4j for clause retrieval, document comparison, relationship analysis, regulatory research, and context-aware summarization.

Engineered PySpark and Dataproc pipelines to cleanse and reconcile governed data from Snowflake and enterprise sources, publishing curated datasets to BigQuery with data-quality checks and auditable validation workflows.

Developed production ML models using scikit-learn, XGBoost, and LightGBM for transaction anomaly detection, behavioral risk scoring, and predictive analytics, applying feature engineering and validation across high-volume financial datasets.

Built near-real-time feature pipelines using Amazon MSK/Kinesis, Spark, and S3, reducing alert-processing latency while supporting AML monitoring, anomaly detection, and risk-based alert prioritization.

Applied SR 11-7-aligned model-risk governance with SHAP explainability, documented assumptions and limitations, reproducibility, validation support, approval gates, and ongoing production monitoring.

Implemented end-to-end MLOps using SageMaker, MLflow, SageMaker Pipelines, Docker, EKS, ECR, and Terraform, reducing model release cycles through automated registration, validation, deployment, rollback, and retraining.

Established production observability using SageMaker Model Monitor, CloudWatch, Prometheus, and Grafana, tracking drift, prediction distributions, model degradation, latency, infrastructure health, lineage, and audit evidence.

Environment: Python, SQL, PySpark, Apache Spark, Scikit-learn, XGBoost, LightGBM, FastAPI, Pydantic, React, TypeScript, Go, LangChain, LangGraph, MCP, A2A, RAG, Vertex AI Vector Search, Dataproc, Cloud Storage, SageMaker, SageMaker Pipelines, Terraform, SageMaker Model Monitor, CloudWatch, Prometheus, Grafana, SHAP.

Client: Centene Healthcare Corporation– St Louis, MO Sept 2023 – Dec 2024

Role: Senior Data & AI/ML Engineer

Overview:

Worked on two healthcare AI initiatives covering multimodal document and image intelligence and clinical NLP. Developed deep-learning, computer-vision, and transformer models using PyTorch, TensorFlow, Hugging Face, BioBERT, ClinicalBERT, CNNs, and Vision Transformers, with AWS-based pipelines for data processing, GPU training, inference, monitoring, and secure deployment.

Key Responsibilities:

Built a multimodal healthcare AI platform using Python, PyTorch, TensorFlow, and Hugging Face to process EHR/EMR records, medical images, clinical notes, claims, and member data, reduced manual document-review effort.

Built CNN, transfer-learning, and Vision Transformer models on NVIDIA GPU-backed SageMaker infrastructure using CUDA/cuDNN acceleration and mixed-precision training, improved validation performance over baseline models.

Developed document-intelligence and medical-image pipelines using Amazon Textract, Optical Character Recognition (OCR), OpenCV, and transformer models to extract clinical attributes, preprocess scanned records.

Developed Streamlit interfaces and FastAPI microservices for ML/CV predictions, integrating API Gateway, Aurora PostgreSQL for relational data, MongoDB for semi-structured metadata, and S3 for model artifacts.

Engineered preprocessing pipelines using S3, AWS Glue, EMR, PySpark, Athena, Python, SQL, and Bash, separating CPU-intensive ETL from GPU training workloads for efficient model-development workflows.

Productionized deep-learning models using SageMaker Pipelines, MLflow, Docker, ECR, EKS GPU node groups, and NVIDIA Triton Inference Server, enabling GPU-aware scheduling, batching, autoscaling, model versioning and rollback.

Secured and monitored workloads using IAM/RBAC, KMS, Secrets Manager, private VPC connectivity, CloudWatch, and CloudTrail, maintaining 99.9% service availability while supporting PHI protection and HIPAA-aligned governance.

Worked on a Clinical NLP platform to analyze EHR notes, care-management documentation, claims narratives, and provider records for clinical entity extraction and document classification.

Built healthcare transformer models using BioBERT, ClinicalBERT, BERT, RoBERTa, PyTorch, and Hugging Face, leveraging GPU-accelerated fine-tuning and mixed precision to improved entity-recognition over baseline models.

Developed NER and medical concept-extraction pipelines to normalize clinical language, map conditions to ICD-10 terminology, and convert free-text documentation into structured ML-ready features with entity-level precision.

Engineered scalable NLP pipelines using AWS Glue, PySpark, S3, EMR, Athena, SageMaker, Python, and SQL, combining claims, eligibility, member, provider, encounter, and care-management datasets.

Productionized ClinicalBERT and BioBERT models using SageMaker, Docker, EKS, NVIDIA Triton, and ONNX Runtime, tuning CPU/GPU instance profiles, batching, concurrency, and autoscaling to reduce inference latency by ~20%.

Established healthcare-model governance using precision, recall, F1, SHAP/LIME, PHI masking, de-identification, drift monitoring, and CloudWatch alerts for production validation and observability.

Collaborated with clinical SMEs, data engineers, MLOps/platform teams, security, and compliance stakeholders to validate healthcare data, define model acceptance criteria, and promote HIPAA-aligned AI solutions into production.

Worked in an Agile/Scrum delivery environment using Jira and Confluence for sprint planning, backlog refinement, stand-ups, technical documentation, code reviews, release coordination, and production issue tracking.

Environment: Python, SQL, PyTorch, TensorFlow, Hugging Face, BioBERT, ClinicalBERT, CNN, ViT, OpenCV, NVIDIA CUDA/cuDNN, Triton, SageMaker, S3, Glue, EMR, PySpark, Athena, MLflow, Docker, ECR, EKS, CloudWatch, IAM, KMS, CloudTrail.

Client: State of Colorado – Denver, CO Feb 2021 – Aug 2023

Role: Senior Data & Machine Learning Engineer

Overview:

Worked for AWS-based Predictive Analytics & Machine Learning Platform focused on scalable data pipelines, feature engineering, model training, deployment, and monitoring, and an Azure-based Machine Learning & MLOps Modernization Platform focused on automated ML workflows, model lifecycle management, containerized inference, and CI/CD. Leveraged Python, SQL, PySpark, Scikit-learn, XGBoost, SageMaker, Azure ML, Databricks, MLflow, Docker, and Terraform to deliver reliable production-grade ML solutions.

Key Responsibilities:

Developed production classification, regression, and forecasting models using Python, Scikit-learn, XGBoost, LightGBM, Pandas, and NumPy, improving predictive performance over baseline models.

Built reusable feature-engineering and model-training pipelines covering imputation, encoding, scaling, feature selection, class balancing, cross-validation, and hyperparameter tuning.

Engineered scalable ETL and feature-processing pipelines using PySpark, AWS Glue, SQL, and Amazon S3, improving data preparation throughput by approximately 25% for model training and batch-scoring workloads.

Trained, versioned, and deployed machine-learning models using Amazon SageMaker, EC2, ECR, MLflow, Docker, and REST APIs, shortening repeatable model deployment cycles.

Implemented model monitoring and data-quality controls using Amazon CloudWatch and Python validation frameworks to detect schema changes, feature drift, prediction anomalies, and model degradation.

Designed end-to-end ML workflows using Azure Machine Learning, Azure Databricks, ADLS, Python, SQL, and PySpark, automating ingestion, preprocessing, training, validation, and scoring across development and production environments.

Built Power BI/Streamlit interfaces backed by FastAPI REST services, integrating predictive ML outputs with Azure SQL and ADLS Gen2 to improve analytics access for downstream analytics and reporting.

Established experiment tracking and model lifecycle management using Azure ML and MLflow, capturing parameters, metrics, artifacts, and model versions, for reproducible validation and deployment.

Containerized predictive services using Docker and FastAPI, exposing REST endpoints for downstream applications and supporting health checks, scalable deployments, and versioned releases with reliable production integration.

Built automated CI/CD and MLOps pipelines using Azure DevOps, Git, Docker, Terraform, and validation gates, reducing manual release steps while improving deployment consistency and rollback readiness.

Partnered with data engineers, analysts, application teams, and stakeholders in an Agile/Jira environment, supporting sprint planning, code reviews, model validation, troubleshooting, documentation, and production releases.

Environment: Python, SQL, Pandas, NumPy, Scikit-learn, XGBoost, LightGBM, PySpark, Apache Spark, MLflow, AWS SageMaker, S3, EC2, ECR, Glue, CloudWatch, Azure Machine Learning, Azure Databricks, ADLS Gen2, Azure SQL, Azure DevOps, Docker, FastAPI, Power BI, Streamlit, Terraform, Git, Jenkins, Jira.

Client: T-Mobile – Bellevue, WA Oct 2019 – Jan 2021

Role: Senior Data Scientist

Overview:

Built a GCP-based Data Science solutions using customer usage, billing, service, and data. Applied predictive modeling, segmentation, forecasting, anomaly detection, and experimentation with Python, SQL, PySpark, BigQuery, Scikit-learn, XGBoost, and GCP services.

Key Responsibilities:

Developed production churn, propensity, and customer-risk models using Python, Scikit-learn, XGBoost, Random Forest, and Logistic Regression, improving predictive lift by approximately 12–15% over baseline models.

Performed large-scale EDA, statistical analysis, feature engineering, and customer segmentation using Pandas, NumPy, SQL, PySpark, BigQuery, and K-Means across usage, billing, service, and network-performance datasets.

Built distributed ETL and feature-processing pipelines using Google Cloud Storage, Dataproc, PySpark, Dataflow, and BigQuery, reducing recurring model-data preparation time.

Designed time-series forecasting and anomaly-detection models for customer usage and network KPI trends, enabling earlier identification of abnormal patterns and improved analytical turnaround.

Applied cross-validation, class-imbalance handling, hyperparameter tuning, feature selection, SHAP explainability, and ROC-AUC/F1/precision-recall evaluation to improve model stability, interpretability, and production readiness.

Conducted A/B testing, hypothesis testing, confidence-interval analysis, and campaign-response measurement to validate retention strategies and translate statistically significant findings into actionable business recommendations.

Trained and deployed batch and online scoring workflows using Google Cloud AI Platform, Docker, Flask/REST APIs, and GKE, while monitoring prediction distributions, data quality, and model drift across production scoring cycles.

Built Tableau/Power BI dashboards on BigQuery and partnered with data engineers, network teams, analysts, and stakeholders in an Agile/Jira environment to communicate insights, validate models, troubleshoot issues, and support releases.

Environment: Python, SQL, Pandas, NumPy, Scikit-learn, XGBoost, Random Forest, K-Means, PySpark, Apache Spark, BigQuery, Google Cloud Storage, Dataproc, Dataflow, AI Platform, GKE, Docker, Flask, REST APIs, SHAP, Tableau, Power BI, Git, Jenkins, Jira.

Client: Amazon – Hyderabad, India Oct 2016 – July 2019

Role: Full Stack Python Developer – Data

Overview:

Built and supported AWS-based retail applications combining Python, FastAPI, Flask backend services, React-based user interfaces, REST APIs, relational databases, and ETL/data-processing workflows. Worked with AWS EC2, S3, RDS, Lambda, Linux, SQL, Pandas, Tableau, KPI dashboards, and ad hoc reporting to deliver scalable applications, automate backend processes, and improve access to sales, inventory, product, and operational insights.

Key Responsibilities:

Developed full-stack web applications using Python, Django, Flask, JavaScript, React, HTML5, CSS3, and Bootstrap, building responsive interfaces and reusable components for internal retail and data-driven applications.

Designed Python, Django, Flask REST APIs and backend services, integrating relational databases, AWS services, and existing Java-based enterprise services to support retail application and data-processing workflows.

Built and optimized relational data models using PostgreSQL, MySQL and Amazon RDS, writing complex SQL queries, joins, stored procedures, and indexing strategies that improved query performance.

Developed ETL and data-processing workflows using Python, Pandas, NumPy, SQL, and Amazon S3 to ingest, transform, validate, and prepare product, inventory, sales, and operational datasets for applications and analytics.

Created Tableau dashboards, KPI reports, and ad hoc analyses using SQL and curated datasets to track sales, inventory, product, and operational performance, reducing recurring manual reporting effort by approximately 20%.

Deployed and supported applications in Linux-based AWS environments using EC2, S3, RDS, Lambda, IAM, ELB, and CloudWatch, performing shell-based troubleshooting, log analysis, monitoring, and production support.

Automated backend processing and CI/CD workflows using Python, AWS Lambda, Linux cron jobs, Git, Jenkins, Docker, and automated testing, reducing repetitive processing and improving release consistency across environments.

Collaborated with product owners, QA engineers, data teams, and developers in an Agile/Scrum and Jira environment, supporting requirements, sprint planning, code reviews, testing, defect resolution, documentation, and production releases.

Environment: Python, Django, Flask, Django REST Framework, JavaScript, React, HTML5, CSS3, Bootstrap, REST APIs, Pandas, NumPy, SQL, PostgreSQL, MySQL, Amazon RDS, S3, EC2, Lambda, IAM, ELB, CloudWatch, Tableau, KPI Reporting, Ad Hoc Reporting, Linux, Shell Scripting, Docker, Git, Jenkins, Jira, Agile/Scrum.

CERTIFICATIONS

AWS Certified Solutions Architect – Associate

Microsoft Certified: Azure AI Engineer Associate

Google Cloud Professional ML Engineer.

NVIDIA-Certified Professional: Generative AI, LLMs.

HashiCorp Certified: Terraform Associate

EDUCATION

Bachelor’s in computer science, GLA University

Contact this candidate