Kaleb Thompson

Experience

1. Principal Architect & Head of Data Engineering · Confidential

   2026 – Present

   Leading platform architecture and the data-engineering function. (Employer confidential for now.)

   Architecture Data Engineering Leadership

2. Principal AI Architect · ONE Gas

   2025 – 2026

   Tulsa, OK

   Sole AI and data architect for the enterprise, reporting to the Chief Data & AI Officer. Architecture lead across data and AI for five business units, 20+ AI initiatives, and a $1–2M annual portfolio.

   • Architected a 40-person, three-pod data & AI organization — data science, data engineering, and data product management — that consolidated under the Chief Data & AI Officer.
   • Built an autonomous LLM agent (Claude / GPT-4) that refactors SQL Server stored procedures, views, and T-SQL into Snowflake SQL + dbt models — 100+ objects at 85%+ accuracy, cutting per-object migration from weeks to hours.
   • Shipped production GenAI: a Safety Copilot scaled from 300 to 2,000+ field users and a legislative risk-analysis tool that cut analysis time 70%, plus enterprise RAG over SharePoint and SQL sources.
   • Built a real-time mixed-integer-programming dispatch optimizer for 200+ field technicians, improving emergency SLA from ~50% to 67%+.
   • Replaced a $250K/yr Gartner advisory product with a custom internal AI intake & innovation platform (React, Tailwind, Valkey, Argo CI/CD, GHCR).
   • Led a $200K cloud-AI vendor evaluation (Azure AI Foundry vs. AWS Bedrock) and set enterprise reference architectures for LLM orchestration, agents, and RAG.
   Applied AI LLMs / RAG MLOps Architecture

3. Head of Enterprise Data & Analytics (Senior Manager) · ONE Gas

   2022 – 2025

   Tulsa, OK

   Owned the company's data strategy, engineering, and science programs across all business lines. Grew the team to 12–14 and managed a $3M+ budget.

   • Built the Snowflake + dbt platform from scratch with CI/CD, Git workflows, and DataOps adopted as the enterprise baseline — 90% improvement in analytical workload execution vs. on-prem SQL Server.
   • Replaced batch ETL with sub-second CDC streaming from Oracle and SQL Server into Snowflake via Qlik Replicate, saving ~$300K/yr.
   • Delivered 30+ enterprise data products and stood up a Power BI Center of Excellence that cut time-to-insight 60%.
   • Reclassified Snowflake spend as Direct Materials to enable CapEx treatment of development costs (OpEx → CapEx).
   Data Platform Snowflake / dbt Leadership Governance

4. Manager, Data Science & Analytics · ONE Gas

   2019 – 2022

   Tulsa, OK

   Led the data-science function and its production ML portfolio; built and led a cross-functional team of engineers and scientists.

   • Built RADAR, a pipeline-strike prediction model (Python, scikit-learn, MLflow, 811 tickets + GIS) that outperformed a $1M/yr vendor at under $100K — leading to a new Damage Prevention department.
   • Deployed Azure Databricks as the company's first cloud ML environment.
   • Shipped Payment Pathways, an ML model scoring bill-pay risk to improve collections recovery.
   Machine Learning MLOps Optimization

5. Lead Data Scientist / Data Engineer · ONE Gas

   2015 – 2019

   Tulsa, OK

   Founded the organization's data-science practice and built its first cloud ML capability.

   • Shipped a meter-read exception classifier that reduced manual review 60% (~$40K/yr), rolled out humans-in-the-loop to earn operator trust.
   • Built predictive capital-project estimators that reduced budgeting variance 10–20%.
   • Delivered field-technician route optimization saving 500+ hours/yr.
   Machine Learning Python Azure Databricks

6. Data Analyst · NRG Energy (Direct Energy)

   2014 – 2015

   Tulsa, OK

   Reconciled billing and revenue data across 20+ utility partners.

   • Recovered $1M+ in missed revenue across 20+ utility partner datasets via SQL reconciliation.
   • Led the transition of transaction-management operations from Toronto to Tulsa.
   SQL Analytics

Skills

Education

Speaking & Leadership

Languages