Dallas, Texas
Senior AI/ML Engineering Architect at New York Life, leading GenAI application delivery, Databricks modernization, and enterprise AI search. Before that, AI/ML Solutions Architect at Ollion (2nd Watch) and Rackspace, and Senior Solutions Architect with AWS Professional Services. I build production LLM, RAG, and agentic systems for financial services, healthcare, insurance, and retail.
Enterprise delivery, from Oracle high-availability databases to production generative AI
Users served by enterprise generative AI initiatives at 99.9% uptime
Models shipped to production on the Data as a Service product I built at 2nd Watch
Professional certifications across AWS, Databricks, Snowflake, Google Cloud, Kubernetes, and Oracle
LLM applications, retrieval-augmented generation, and multi-agent systems with evals, guardrails, and human review designed in from the start, so they survive real traffic and a security review.
Now: ClientMine and ProspectMine, New York Life's GenAI client-intelligence apps. Agent projects on GitHub
Lakehouse modernization on Databricks and Snowflake: Redshift migrations, medallion pipelines, Databricks Apps and Lakebase, Unity Catalog, and the identity and networking work that gets them through security review.
Recent: PRISM migration to Databricks Apps and the Redshift-to-Databricks program at New York Life.
Fine-tuning, model serving with vLLM and Triton on GPU fleets, prompt and eval registries, and AI governance mapped to NIST AI RMF, the EU AI Act, and SR 11-7 for regulated industries.
Built: QLoRA fine-tuning and vLLM serving on AWS, cutting inference cost 60% with model selection and caching. See the repos
Mar 2022 - Present
New York Life Insurance (TDAV)
Lead architecture, security review board readiness, and implementation planning for ClientMine 3.0 and ProspectMine, New York Life's GenAI client-intelligence applications, including the production deployment that put ClientMine live for the field. Production RAG applications handle 100K+ daily queries; multi-agent systems on LangGraph cut customer-service response time 75%. Led the PRISM migration to Databricks Apps, re-architected the IBF platform on Databricks, and architected the enterprise Glean AI search platform on AWS for a pilot spanning 9 business units and 8 data connectors.
Sep 2021 - Mar 2022
Ollion (formerly 2nd Watch)
Led a customer's migration from an on-prem warehouse and Hadoop estate to a cloud lakehouse with Terraform and orchestrated pipelines, cutting batch runtime from about 6 hours to 45 minutes and infrastructure cost about 30%. Built the Data as a Service product that let the AI/ML team ship 80+ models to production, and presented it at AWS re:Invent. Converted PyTorch models to TensorRT and ONNX runtimes serving 10,000+ QPS.
Mar 2020 - Aug 2021
Rackspace Technology
Built and productionized a demand-forecasting pipeline for a multi-location retailer, working directly with merchandising teams to fold forecasts into replenishment decisions. Designed and deployed a customer-churn model on AWS that flagged at-risk accounts ahead of cancellation, then helped the client's product team build retention programs from its output.
Jun 2019 - Mar 2020
ActiveSkyGate LLC
Delivered end-to-end AI/ML solutions on Amazon SageMaker. Shipped TensorFlow Serving recommendation models that lifted click-through 12%, migrated scikit-learn pipelines to TensorFlow/Keras for 45% more throughput, and automated quantization and pruning that cut model memory 60% for mobile deployment.
Apr 2017 - May 2019
Amazon Web Services (Professional Services)
Built end-to-end MLOps pipelines on Amazon EKS and Redshift, cutting model deployment time from weeks to hours. Created production FastAPI services serving 50+ AI models with auto-scaling, and built real-time ingestion with Apache Kafka for ML feature stores.
Mar 2016 - Mar 2017
Oracle
Led customer operations and projects for Database High Availability products serving 400,000+ customers. Deployed, tuned, and supported ODA and Exadata systems running Oracle 11g and 12c.
Jan 2015 - Mar 2016
SoundHound AI (Amelia, formerly IPsoft)
Designed conversational AI and chatbot frameworks, built NLP pipelines for document understanding and information extraction, and automated model training and evaluation in Python.
Jan 2013 - Dec 2014
Sodexo
Developed computer vision models for image classification and object detection in TensorFlow and PyTorch, and built processing pipelines for large image and text datasets.
Jan 2010 - Dec 2012
Biodenta North America
Built enterprise applications in Python and SQL, ETL systems for data warehousing and BI, and API integrations.
Data as a Service, presented at AWS re:Invent in Las Vegas from the 2nd Watch booth: how a data product let an AI/ML team ship 80+ models to production, built on S3, Lake Formation, Glue, Athena, EMR, Redshift, Aurora, and OpenSearch.
Scalable data lakes on AWS as an internal product: one governed path from raw data to models in production.
AWS re:Invent · 2nd Watch booth
A 16-module program from prompt engineering and RAG through agent architectures, MCP, multi-agent systems, and enterprise deployment on AWS, Azure, and GCP.
Curriculum and instruction · Next cohort January 2027
AI risk and governance frameworks for analysts entering the field: NIST AI RMF, the EU AI Act, and SR 11-7 in practice.
Program design and instruction
Open-source builds, deployed on AWS with Terraform and torn down afterwards to control cost. Every repo ships with real models and real infrastructure, no mocks.
Telecom customer-care voice agent on Amazon Connect and Lex V2 with a LangGraph agent on Bedrock, guardrails, PII redaction, and RAG over policy documents. Tested end-to-end on a live toll-free number.
Amazon Connect · Bedrock · LangGraph · Terraform
An agent that lives in Slack and searches, reads, and summarizes Google Drive. LangGraph ReAct agent with LangSmith tracing, deployed as a container on ECS Fargate; v2 adds a multi-agent supervisor team.
LangGraph · Bedrock · LangSmith · ECS Fargate
QLoRA fine-tuning of Mistral-7B to triage insurance support tickets into category, priority, and a draft reply, trained on a Terraform-provisioned GPU instance with an eval harness.
QLoRA · Mistral-7B · EC2 GPU · Terraform
Production vLLM serving of Llama-3.1-8B-Instruct as an insurance Q&A assistant, with a gateway, load tests, Prometheus monitoring, and a live serving dashboard on autoscaling GPU instances.
vLLM · Llama 3.1 · AWS · Terraform
Energy anomaly detection from Databricks medallion pipelines to an XGBoost model in MLflow, CI quality gates, FastAPI serving, Evidently drift monitoring, and a NIST AI RMF model card.
Databricks · MLflow · XGBoost · Evidently AI
Agentic RAG, multi-agent eval harnesses, MCP servers, GPU inference experiments, and the Terraform behind each one.
github.com/iamdibo365 →
More builds
Certifications
AWS Machine Learning - Specialty
AWS Machine Learning Engineer - Associate
AWS AI Practitioner
AWS Solutions Architect - Professional
AWS Solutions Architect - Associate
AWS Developer - Associate
AWS SysOps Administrator - Associate
AWS Cloud Practitioner
Snowflake SnowPro Core
Google Cloud Professional Data Engineer
CNCF Certified Kubernetes Administrator (CKA)
HashiCorp Terraform Associate
Oracle Database 12c Administrator Professional
Oracle Database 11g Administrator Professional
Oracle Database 11g Administrator AssociateIn progress
Education
New Mexico Highlands University
2010 - 2013 · GPA 3.91
University of Buea, Cameroon
2006 - 2009 · GPA 3.75
Enterprise AI/ML architecture, LLMOps, and AI governance.