Scalable solutions
Systems that keep working after the launch meeting: scheduled, tested, and monitored. My healthcare dashboard refreshes itself monthly and runs 486 automated tests on every change.
Monthly refresh · 486 tests
Solution architect · Business operations & AI
I design the systems that turn business questions into decisions.
Ten years in operations, the last eight in revenue operations, deal desk and analytics. I map how the business really runs, design the system that fits it, and now build multi-agent AI systems with evidence behind every answer.
Positioning
Operations first. Then the AI.
Most AI projects fail on the operations side: bad data, unclear owners, no one checking the output. I come from that side. I've run deal desks, rebuilt sales processes through four acquisitions, and put the numbers in front of a board. Now I design AI systems with that same discipline: clean data first, math done by code, models used only for judgment, and a check before anything reaches a decision-maker.
Data-driven strategy
Forecasts and market sizing that leadership actually used: a demand forecast that improved accuracy by 5% and saved $1M a year, and TAM analysis delivered to investors.
$1M/yr saved · +5% forecast accuracy
Process design
I map current and future state before anything changes. At PG Forsta I redesigned the path from sales input to proposal, contracting, and approval, and separately streamlined the sales cycle to shorten it.
Sales input → proposal → contract → approval
Change management
New systems only pay off when people use them the same way. I train teams on new processes so the data they produce stays clean, and I've carried operations through four acquisitions.
4 acquisitions
AI agent systems
Multi-agent systems designed around real business questions, with models chosen by measured tests, not reputation, and guardrails that stop unsupported claims from publishing.
6 models tested · 2 providers
- Eleos HealthGTM Operations Analyst
Customer Success dashboards (client risk, NPS, renewals, NRR) and dataflow reconciliation, built primarily in Streamlit.
2026–Present - PG Forsta (acquired by Qualtrics)Senior Manager, Deal Desk
Owned Deal Desk for $10M+ negotiations, ran GTM strategy and Salesforce rollout, and guided the team through four acquisitions.
2023–2026 - PG ForstaSenior Operations Analyst
Restructured territories and reporting infrastructure; built predictive models for forecasting, risk, and retention.
2021–2023 - Jubilant HollisterStierSales Operations Analyst
Built a demand-planning model that improved forecast accuracy 5% and owned Salesforce CRM data governance.
2018–2021
Skills
The toolkit behind the work.
Revenue Operations
- Deal Desk
- Salesforce
- Sales Compensation Design
- Territory & Pipeline Management
- RFP / Pricing & Packaging
Data & Reporting
- Streamlit
- Power BI
- Tableau / Tableau Prep
- SQL / SOQL
- DBT
- Amazon Athena & S3
Statistical Modeling
- Random Forest
- Holt-Winters Forecasting
- ANOVA / T-tests
Emerging: Agentic AI
- LLM Agent Orchestration
- GitHub Actions Pipelines
- Claude & OpenAI APIs
How I work
Map, design, build with AI, govern, enable.
Map
Find the real question and the current process, owners, and data before proposing anything.
Design
Architect the system: what code computes, what a model judges, what a person approves.
Build with AI
Direct AI coding agents from written specs, with tests as the definition of done.
Govern
Guardrails, evaluations, and monitoring, so the output can be trusted without someone re-checking it.
Enable
Train the people who use it, and hand over docs they can run it from.
Watch the intro
60–90 seconds on how I work.
Silent-first: every point is on screen as text, with captions for the voiceover.
The intro video is coming soon.
Featured project · Healthcare Intelligence Dashboard
A live example: an executive dashboard run by a team of AI agents.
A health plan's leadership needs to know what changed this month across markets, costs, providers, Medicare, Medicaid, and policy, and why. I designed a team of 13 AI agents that reads 32 public federal data sources every month and turns them into ranked findings, each with its evidence one click away.
13
AI agents: an orchestrator, 10 specialists, 2 infrastructure
32
Public federal data sources
6
Models tested from two providers
Monthly
Automatic refresh, 1st of each month
What I did. I framed the executive questions, designed the agents and the evidence model, set the guardrails, directed AI coding agents to build it, and chose the models by testing six of them on the same benchmark.
What it costs. Each job runs on the model picked by measured cost and quality in a six-model benchmark, not by prestige.
Code computes every number. Models only pick and explain real findings, and a grounding check drops any number the data doesn't support before anything publishes.
The guardrail · Healthcare Intelligence Dashboard
Next step
Have a process that should run itself?
Tell me the decision you need to make and where the data lives. I'll tell you honestly whether an AI system is the right answer.