Kurt Brischke

Deriva Energy · Brookfield Renewable · Charlotte, NC

Systems builder across three industries: family ERP in Nicaragua, architect of the Databricks Lakehouse behind a 5.4 GW wind and solar fleet at Brookfield Renewable, product owner on Lighthouse at Duke Energy, and a vault-native personal AI operating system run solo. 15+ years inside the asset, not pitching to it. Where AI work today needs judgment, design, and first-principles ownership more than raw engineering horsepower, I span all four: code, AI leverage, business, and hardware.

Save Contact Systems I Run

By the Numbers

$20M+

Career-Wide Realized Impact Across AI & Analytics Initiatives

3.1 GW

Wind Fleet · 1,300 Turbines Across 21 Sites

2.3 GW

Solar Fleet · 1,700 Inverters Across 67 Sites

1.6M

PI/SCADA Sensor Points Across the Fleet · 10-Minute Micro-Batch Cadence

Patent

Co-Inventor, Descriptive Analytics (SMART)

EPRI

Award for Innovation in Renewable Analytics

Systems I Run

Private systems, built solo and operated daily. Not demos and not portfolio pieces: each one runs the same discipline I apply at fleet scale. A compiled data spine, generated decision surfaces, and verification before anything publishes.

Market Operations

The Finance Desk

A private market-operations desk: roughly 235 self-contained generated pages compiled from a portfolio spine, rebuilt six times a day from live market data. Six question-shaped tabs (Today, Atlas, Horizon, Tactics, Stocks, Reports), each answering one question, with per-position drill-through dossiers underneath.

One action engine, one producer. Every candidate action is scored value × pressure × confidence. Confidence is allowed to be zero, and zero routes the item to a data-debt lane instead of dressing it up as a guess.
Atomic build-verify-publish. A 35-assertion regression harness runs on every build. Any failure is fatal: nothing publishes, and the previous build stays live.
Private by construction. Served loopback-only through a Cloudflare tunnel behind Cloudflare Access. Zero external fetches on any page.

Finance Desk: per-position drill-through dossier

Figures shown are display-scaled demo numbers, not live positions.

Static Generation Regression Harness Cloudflare Access Zero External Fetches

Personal Health

The Health System

An internal system, not a website. A whole-body osteopathic operating frame sits on top of the data: static protocol documents generated from lab panels (Function Health), Eight Sleep data, and a home sleep study (HSAT), compiled into monthly protocol plans. A single advisor with four modes covers the nervous system, relationships, meaning, and the body, so the protocols and the person running them stay in one conversation.

Health System: generated monthly protocol document

Sample protocol document. Values shown are illustrative.

Generated Protocols Lab-Panel Driven Eight Sleep + HSAT Monthly Plans

Personal AI Operating System

A vault-native operating system for one person's life and work, running on plain markdown, git, and a compiler. The arc across every piece of it: build, measure, simplify.

Five specialist advisors, one voice. One advisor per life domain (money, career, technology, logistics, and an integrated mind-and-body advisor), invoked as skills with cross-domain escalation rules. Specialist lenses, not separate agents: the system adopts a perspective and answers as one.

Skills framework. 30+ modular skills for document generation, research workflows, financial data collection, and orchestration, each with its own trigger and routing rules. Judgment-as-code.

Vault-native memory. Built a 26-tool MCP semantic-memory server (PostgreSQL + pgvector, hybrid semantic and full-text retrieval over 4,400+ memories), then deliberately merged it into a simpler vault-native design: dated markdown, grep-able, versioned in git. The server was the measurement; the vault won.

Scheduled automation with receipts. Automation that leaves evidence: every scheduled run writes a verifiable receipt, and a weekly drift audit hunts stale facts across the whole estate.

Canonical-truth compilation. Attested facts live as individual source files and compile into a generated registry. Every surface reads the registry; conflicts resolve at the fact, never at the page.

Build, measure, simplify. The simplification is itself the credential: knowing what to keep is harder than knowing what to build.

Foundations First / How I Ship

AI is the multiplier. The four layers below are what it multiplies. Each one extends what the layer beneath it makes possible. Get them right and AI compounds across the business. Skip them and AI-first collapses on first contact with reality.

Business Operations DNA

IT Systems + Data Discipline

Product Judgment + Design

AI Products That Ship

The compounding shows up when all four layers sit on top of each other. The four principles below are how I work that compounding in practice.

First principles over status quo. Every system I touch, whether a 3,000-person ERP in Nicaragua, the Lakehouse behind a 5.4 GW fleet, or a personal AI operating system, starts from the physics of the problem and the shape of the data, not the latest model or the existing tooling.

Judgment and design are the durable work. When 1 to 2 engineers plus AI compound to what used to take 15, the bottleneck shifts to product ownership, scoping, and knowing what not to build. I product-owned 6 data scientists on Lighthouse at Duke. Now I ship solo at similar velocity.

Data sensitivity is a feature, not a burden. What moves, what stays, who sees it. Two years as enterprise Data Steward and migrating a 1.6M-point PI/SCADA sensor estate off-prem taught me the calls AI-first organizations are about to face at scale.

Real-world constraints drive architecture. SCADA latency, 300-foot tower climbs, control-logic timing, sensor drift. The hardware teaches you what clean abstractions cannot, and it shows up in every design decision that has to survive contact with a physical asset.

Enterprise Experience

Senior Manager, Data Science & Analytics · AI Center of Excellence

Deriva Energy (Brookfield Renewable) · 2021 – Present

Databricks Lakehouse Architect. Architect of the Lakehouse design and deployment: Delta Lake, Unity Catalog, medallion architecture, 10-minute micro-batch streaming. Consolidates operating data from 3.1 GW of wind (1,300 turbines, 21 sites) and 2.3 GW of solar (1,700 inverters, 67 sites), a 1.6M-point PI/SCADA sensor estate. $20M+ in career-wide realized impact across initiatives.
Chilton Manual for Wind Turbines. COO-sponsored LLM diagnostic copilot mapping SCADA faults to resolution data across 90+ OEM manuals and 100,000+ fault events. Projected $3.6M to $5.0M per year steady-state. Led architecture POC validation with IT, operations, and data-science stakeholders before full build.
Production ML at Scale. Directed multi-year McKinsey collaboration deploying 22 component-level failure forecasting models. Reduced unplanned downtime 18 to 25%. Prevented 3+ catastrophic failures ($1.5M+ each).
Product Owner across 3 agile teams (data science, analytics engineering, operations). Brookfield AI Champion leading 14+ AI use cases at Deriva within Brookfield's 15-portfolio-company, 7-country AI Value Creation Office, driving toward the $10M 2026 AI value target.
$7.7M saved through LTSA compliance and vendor LD recovery (2021–2026). Negotiated multi-year platform contracts with Seeq and Onyx.

Previous: Duke Energy Renewables, American Apparel (Director of Capacity Planning), Ashley Furniture, Pinehurst Manufacturing

15+ years across renewable energy, manufacturing, retail · 6 years US + Nicaragua (bilingual) · Microsoft Professional Degree in Data Science

Speaking & Recognition

EPRI Award. Innovation in Renewable Analytics.

Patent Co-Inventor. Descriptive Analytics: SMART (Solar Monitoring and Reliability Tool).

Brookfield AI Summit. Inverter PdM with AWS (Toronto, 2025).

Seeq Conference Presenter. Bat Curtailment Analytics.

McKinsey Collaboration. Multi-year ML engagement for turbine failure forecasting.

Education. MBA (UNC Charlotte) · B.S. Mgmt & Acct (USC, Full Scholarship) · Studied abroad: Italy, Spain

LLM Architecture RAG Systems Agentic AI Claude API / MCP Databricks Delta Lake / Unity Catalog AWS SCADA / PI Python + SQL Vector Embeddings Power BI Product Ownership Predictive Maintenance Data Governance Executive Translation Spanish (Proficient)

Get in Touch

Save Contact

A current resume is available on request.