💻 LinkedIn · 📍 Ohio / Remote

I uncover the questions no one thought to ask and build the systems to answer them.

I build data and automation systems for messy operational problems — especially where the difficult part is figuring out what the data actually represents before deciding how to model, measure, or automate it.

My work spans source-system investigation, analytics engineering, dimensional and semantic modeling, analytical automation, and independent agentic-systems R&D.

I use agentic systems as a systems-engineering problem rather than treating model capability as sufficient evidence that an agent should be allowed to act.

My current OpenClaw environment operates as an orchestrated multi-actor runtime built around 12 reusable agent skills. Skills encode reusable decision logic alongside failure modes and explicit boundaries for when they should not be used.

I also designed an L0–L5 authority model that separates what an agent can do from what it is authorized to do. Consequential actions are governed through escalation paths and human-override controls rather than relying solely on runtime model judgment.

A provenance and capability audit evaluates state across memory, knowledge, skills, ownership, dependencies, access boundaries, stale references, and configuration drift. Its first full run identified 19 broken cross-references and one missing-ownership regression, which were then converted into additional system-level controls.

The retrieval layer tracks source identity, evidence weighting, confidence, contradictions, and unresolved questions so retrieved information can be evaluated for more than semantic relevance alone.

Real-time NBA data application designed for Even Realities G2 glasses and R1 ring interactions.

flowchart LR
    Scheduler[Scheduled Polling]

    subgraph APIs
        ESPN[ESPN Schedule API]
        NBA[NBA API]
    end

    subgraph Cloudflare
        Worker[Cloudflare Worker]
        Cache[KV Cache]
    end

    subgraph Clients
        Web[Web App]
        G2[Even G2]
        R1[Even R1]
    end

    Scheduler -->|Every 15 sec| Worker
    Worker --> ESPN
    Worker --> NBA
    Worker <--> Cache

    Worker --> Web
    Worker --> G2
    Worker --> R1

The application automatically selects active games, falls back to upcoming schedule data when no game is live, caches responses through Cloudflare KV, and exposes live status and play-by-play data across the web interface and device interactions.

Built a workforce analytics system that unified operational and employee-performance data across Five9, Paylocity, Salesforce, Pioneer, TECSYS, and Snowflake.

flowchart LR
    subgraph Sources
        F[Five9]
        P[Paylocity]
        S[Salesforce]
        PH[Pioneer]
        T[TECSYS]
    end

    subgraph Snowflake
        Raw[Raw Data]
        Events[Unified Events]
        SCD[Historical Employee Dimension]
        Model[Scored Analytical Model]
    end

    subgraph Looker
        Explore[Governed LookML]
        Reporting[Self-Service Analytics]
    end

    Sources --> Raw
    Raw --> Events
    Events --> SCD
    SCD --> Model
    Model --> Explore
    Explore --> Reporting

A major challenge was historical attribution.

Employees changed managers, teams, and cost centers over time, which meant joining historical activity to current employee state produced incorrect reporting.

I built a Type 2 Slowly Changing Dimension with effective-date ranges so work could be attributed according to the employee's organizational state when the work occurred.

I also built event-level attribution logic using SQL window functions to determine which employee actually performed specific fulfillment tasks instead of relying on current or order-level assignment.

The resulting analytics system supported more than 200 employees, eliminated approximately 15 hours per week of manual Excel reporting, and became a common analytical source for workforce-performance reporting.

My current work focuses on analytics and optimization for contact-center operations.

I build reusable analytical frameworks that combine operational, CRM, telephony, contact, booking, and workforce data; investigate source behavior and data-quality issues; develop SQL, Python, and Alteryx workflows; and build Power BI semantic models for operational and financial analysis.

A growing portion of the work is focused on analytical automation: converting recurring investigation and business-review processes into reproducible Python-based systems that identify meaningful movements, validate metrics, and generate structured reporting outputs.

I also support metric standardization, documentation, reproducible query logic, and data-governance practices across analytical work.

Worked across the full analytics stack from source-system discovery through Snowflake modeling, governed LookML, operational analytics, and reporting.

Additional systems included a self-maintaining Snowflake business calendar, real-time SLA monitoring designed to surface revenue at risk before failures compounded, CI/CD workflows through GitHub and CircleCI, and analytics involving HIPAA-regulated PHI and sensitive PII.

Before moving fully into data analytics, I worked extensively with workforce-management and operational systems across forecasting, capacity planning, system configuration, automation, troubleshooting, documentation, and long-term system maintenance.

That background is a major reason I tend to treat data problems as systems problems first.

When a number is wrong, the problem is often not the calculation.

I tend to work backward through the system:

flowchart LR
    S[Source System]
    E[Event Semantics]
    G[Grain]
    I[Identity + Time]
    M[Data Model]
    K[Metric Logic]
    V[Validation]
    A[Automation]
    D[Decision Support]

    S --> E
    E --> G
    G --> I
    I --> M
    M --> K
    K --> V
    V --> A
    A --> D

The questions I care about are things like:

• What business event does this record actually represent?
• What is the true grain?
• Which timestamp corresponds to the event being measured?
• Is this current state or historical state?
• Which source is authoritative?
• What did a join add, remove, or duplicate?
• Does the metric remain valid under different filter contexts?
• Is the discrepancy in the calculation, transformation, source data, or underlying operational process?

The goal is not simply to produce a result.

It is to understand the mechanism well enough that the result can be trusted and the solution can be reused.

B.S. Data Science — University of Maryland Global Campus In Progress · Expected 2028

GPA: 4.0

Academic focus: Algorithmic bias, Responsible AI, data governance, and human-centered evaluation of automated systems

Honors: Dean's List · Alpha Sigma Lambda, Tau Chapter — Fall 2026 Inductee