Senior AI/ML Architect with 11+ years shipping production-grade GenAI, agentic AI, and lakehouse data platforms on AWS, Azure, and Databricks across manufacturing, insurance, finance, healthcare, and retail.
I design and ship agentic AI systems, GenAI applications, and large-scale data platforms that real users depend on. Over 11+ years, I’ve moved from analytics to data engineering to ML productionization, and now to agentic AI architecture — combining hands-on engineering with technical leadership.
Currently at Hubbell Incorporated (Cleveland, OH), I architect multi-agent systems on LangGraph, LangChain, and CrewAI, RAG pipelines with Pinecone & Milvus, and MLOps infrastructure on AWS & Databricks. Before that, I led full-stack ML platforms at Allstate, Thomson Reuters, Cybage, and Indium Software.
I hold a Master’s in Business Analytics & Information Systems, a Bachelor’s in Electronics & Communication Engineering, and certifications across AWS, Azure, Databricks, and Python. I’m a player-coach — deep in code, but equally focused on mentorship, architectural reviews, and turning ambiguous business problems into shipped systems.
A pragmatic stack for shipping reliable agentic AI and cloud-native ML platforms.
Architecting multi-agent systems on LangGraph & LangChain, RAG infrastructure on Pinecone & Milvus, and the MLOps platform on AWS + Databricks.
Built a full-stack insurance analytics platform with FastAPI & Vue.js, Kafka pipelines, fraud detection (graph anomaly), and claim severity/frequency models.
Shipped ML models to production on AWS SageMaker with multi-variant A/B testing, time-series forecasting (+18% accuracy), and document classification at 100k+ scale.
Built large-scale AWS ETL pipelines for healthcare claims, EHR/EMR data, and HL7 / X12 / FHIR integrations under HIPAA compliance, processing multi-TB datasets.
Designed 50+ Tableau and Power BI dashboards, automated reporting with Python (cutting manual effort 60%+), and ran A/B testing for marketing analytics.
Real production systems across five domains — not prototypes, not demos.
Architected a stateful multi-agent system on LangGraph with planner–executor patterns, tool use, memory, and self-validation loops — powering autonomous decomposition of complex manufacturing-ops requests.
Built RAG infrastructure with HNSW-indexed Pinecone & Milvus stores at billion-scale, optimized for tool-calling and JSON-mode outputs across enterprise document corpora.
Engineered a real-time fraud-detection engine combining graph anomaly detection (NetworkX) with ensemble ML — surfacing duplicate-claim and staged-accident patterns that directly reduced loss-ratio exposure.
Implemented ARIMA & Prophet pipelines for quarterly revenue and cash-flow forecasting on AWS SageMaker — improving forecast accuracy by 18% over statistical baselines.
Built large-scale ETL pipelines on AWS S3, EMR, and Redshift ingesting HL7, X12, CCD/CDA, and FHIR formats — powering HEDIS, risk-adjustment, and claims analytics under HIPAA controls.
Implemented full model lifecycle tooling using AWS SageMaker Model Registry, MLflow, blue/green deployments on EKS, and CloudWatch-driven drift detection for production resilience.
Open to Lead AI Engineer, AI/ML Architect, and Agentic AI roles. If you’re shipping autonomous AI systems and need a hands-on architect, I’d love to hear from you.