Cory Tillman

Technical Program Manager | AI & Automation Architect

I design scalable operational programs and AI-powered systems that turn chaos into cadence.

I turn chaotic, manual operations into governed, measured, partly-automated systems, and translate operational data into leadership decisions. I span program management, data and root-cause analysis, no-code workflow automation, and lightweight data and AI engineering. My bias is to build mechanisms and measure what matters — achievable targets over aspirational ones.

View My Work Download Resume
300+ Sites Managed 83% Time Reduction 22,357 Checks Delivered
About

Program manager. Systems builder. Data-first operator.

Professional bio

I'm a technical program manager and operations engineer in on-site technology support (field audio-visual and IT) across a large, multi-site enterprise. I turn chaotic, manual operations into governed, measured, partly-automated systems, and translate operational data into leadership decisions.

I'm comfortable spanning program management, data and root-cause analysis, no-code workflow automation, and lightweight data and AI engineering. Whether it's a preventative-maintenance cadence for a distributed equipment fleet, a serverless reporting pipeline, or a human-in-the-loop AI classification agent, I build the mechanism, prove it with data, and hand leadership a decision.

Core capabilities

  • Program & project management — discovery, requirements, scope, RACI, RAID, dependencies, scheduling, governance, change control, executive reporting, closeout.
  • Data analysis & RCA — large operational datasets, SLA analytics, 5 Whys / fishbone, trend reporting with stable taxonomies.
  • Workflow architecture — Asana and Smartsheet: forms, rules engines, cross-sheet logic, dashboards, reusable templates.
  • Data engineering (lightweight) — serverless pipelines (ingest → enrich → validate → visualize), CSV/data wrangling, dashboarding.
  • Applied AI — classification/summarization workflows, prompt architecture, agentic workflows, MCP tooling, safe human-in-the-loop design.
  • Executive communication — narrative-with-numbers reporting, business cases, scenario modeling.

Certifications & learning

Five certifications earned across cloud, work management, and networking — two more in progress.

AWS Certified Cloud Practitioner

CLF-C02 · 2026

Certified ✓

Asana Administrator Certificate

Atlassian-Asana · 2026

Certified ✓

Asana Workflow Specialist

Asana · 2025

Certified ✓

Smartsheet Core App + Builder Path

Smartsheet · 2026

Certified ✓

Full Stack Network Professional

NexGenT · 2021

Certified ✓

AWS Certified AI Practitioner

AIF-C01

In Progress

AWS Solutions Architect Associate

SAA-C03

In Progress

Tools I've built myself

Personal IP — evidence of applied AI and automation, built end to end.

Multi-agent AI framework

Role-specialized agents — architect, auditor, implementer, tester, guardian and more — with explicit handoffs and quality gates.

Custom MCP knowledge server

A durable knowledge/memory server (Python + SQLite) that gives agents structured, persistent context.

File classification & routing utility

Automated classification and routing that turns unstructured file drops into organized, actionable structure.

Reference / formula helper web app

A full-stack helper app for formula patterns and solution design references.

Cert-study apps

Self-built study tools for AWS AI Practitioner and Cloud Practitioner — learning by building.

Systems Portfolio

Seven systems, designed and delivered.

Each turned a chaotic, manual process into a governed, measured mechanism. Select a card for the full story.

Methodology

How I run programs — and how I use AI.

Two repeatable playbooks. One builds governed programs; the other keeps AI accountable.

TPM Methodology

1Discovery
2Requirements
3Build
4Deploy
5Govern
  • Start from the problem, quantified. Define the problem with data before proposing anything.
  • Define success up front. Metrics and targets are set before work starts, not reverse-engineered after.
  • Build mechanisms, not heroics. If it depends on someone remembering, it will fail — encode it in a rule, template, or cadence.
  • Make ownership explicit. Every workstream, action, and risk has one accountable owner.
  • Achievable beats aspirational. A target missed every period trains people to ignore targets.
  • Close the loop. Cadence reporting and lessons learned turn one-off wins into systems.
  • Reversible by design. Pair any significant change with a rollback trigger.

AI Methodology

Data
Clean
AI Analysis
Validate
Human Review
Output

Core belief: AI accelerates the analyst; it does not replace accountability. Every AI output that reaches a decision-maker passes through validation and a human-review gate.

  • Structured classification. Force a JSON schema with a confidence field; low-confidence items route to a human.
  • Summarize-from-structure. Summarize validated structured data, not raw text, so narratives can't invent facts.
  • Extract-then-reason. Extract fields deterministically first, then reason over the extraction to reduce hallucination.
  • Critic pass. A second prompt checks the first output against its constraints.
  • Deterministic scaffolding. Orchestration, I/O, and validation are code; only the reasoning is the model.
  • Approve-to-publish. Nothing leadership-facing ships without human sign-off.
Contact

Let's build something together.

Open to TPM and AI/automation roles, consulting leads, and interesting problems.