Autumn campus plaza with brick buildings and fall foliage

Case Study

UMGC

  • Next.js
  • React
  • TypeScript
  • Azure
Employed on 2024 - Present

Overview

I joined the University Of Maryland Global Campus as Lead Developer on November 18, 2024. Since then the work has been a steady arc of shipping platform products — not a single big bang, but a sequence of modernizations that keep production stable while the stack moves forward.

The through-line: guide a hybrid team maintaining a legacy AngularJS portal while standing up a modern Next.js platform, tighten how code ships, raise quality gates, and — most recently — stand up per-team copilot agent pilots wired into Azure DevOps workflows.

Role
Lead Developer
Scope
Portal modernization, CI/CD, copilot agents
Type
University Of Maryland Global Campus
Year
2024–Present

Colors

Primary
Secondary
Tertiary
Accent

The journey

Products shipped along the way

Five milestones from first month to present — each a shippable increment, not a slide deck.

Next.js platform

Portal modernization

The public portal had to keep serving traffic while the stack moved from AngularJS toward Next.js — contractors and in-house engineers on the same codebase.

Hybrid delivery model: maintain legacy routes, stand up Next.js surfaces, and migrate feature-by-feature without a big-bang cutover.

  • AngularJS maintenance in parallel with Next.js feature work
  • Shared component and routing patterns across legacy and modern layers
  • Contractor coordination with in-house code and design review
Git & CI/CD

Delivery infrastructure

Merge conflicts and slow deploy cycles were costing the team more than feature work.

Architected a Git branch strategy and CI/CD pipeline so releases are predictable, reviewable, and traceable from PR to production.

  • Branch strategy tuned for hybrid in-house and contractor contributions
  • CI/CD pipeline with gated checks before publish
  • Publish success above 88% with low failure rates
Platform layer

Performance & data handling

Perceived speed and maintainable data flows directly affect whether a modernization sticks.

Front-end performance passes and modernized data-handling patterns so the Next.js platform stays responsive under real traffic.

  • Load-time and runtime performance optimization on critical paths
  • Data-fetch and caching patterns aligned with the new stack
  • Production monitoring feedback loop into review gates
Dev experience

Team enablement

Platform work only scales when junior and mid-level developers can ship confidently.

Mentoring, code and design reviews, and a technical wiki that captures architecture decisions and system workflows.

  • Mentoring on front-end architecture, refactoring, and code management
  • Wiki documentation for branch strategy, patterns, and workflows
  • Review culture that keeps accessibility and production readiness non-negotiable
Azure DevOps

Copilot agent program

Teams needed more than a chat box — they needed agents orchestrated around how they already work in Azure DevOps.

A copilot agent starter kit, per-team pilots, a custom CLI, and orchestration layer for Azure DevOps workflows — PRs, pipelines, and repeatable agent setup.

  • Starter kit patterned after agent-orchestration best practices
  • Per-team pilots so adoption stays owned locally
  • CLI to scaffold agents and wire Azure DevOps hooks
  • Orchestration for pipeline and PR workflows — the journey capstone

Starting point

A platform that could not pause for a rewrite

When I joined, the organization needed forward motion on a live portal — without the luxury of a clean-room rebuild.

Legacy stack

AngularJS still carrying production traffic

Critical flows ran on a mature AngularJS codebase while the target architecture moved toward Next.js and modern React patterns.

Hybrid delivery

Multiple contributors, one release train

In-house engineers and contractors needed a shared branch strategy and review bar so merges stayed predictable.

Quality bar

Performance and accessibility could not be afterthoughts

Every release had to meet production readiness — WCAG-aligned reviews, faster loads, and publish pipelines that rarely failed.

How we ship

Modernize without breaking production

Four principles kept the journey coherent from the first sprint to the copilot agent rollout.

Capabilities

Copilot agents built for how teams ship

The capstone product is not a generic chatbot — it is a copilot agent program with orchestration for Azure DevOps and a CLI that makes each team's pilot repeatable.

Copilot Agent Starter

Orchestrated agents for Azure DevOps

Each dev team runs its own pilot from a shared starter kit: agent definitions, workflow routing, and Azure DevOps integration — scaffolded through a CLI so setup stays consistent without central bottlenecks.

  • Per-team copilot agent pilots from a shared starter kit
  • CLI to scaffold agents and wire Azure DevOps workflows
  • Orchestration across PR review and pipeline tasks
  • Publish pipeline holding above 88% success with low failures

Branch strategy

Git workflow designed for hybrid teams so feature branches, reviews, and releases stay predictable.

Release gates

CI/CD checks and production-readiness review before every publish — keeping failure rates low.

Mentoring

Hands-on guidance for junior and mid-level developers on architecture, refactoring, and shipping safely.

Portal modernization

Parallel legacy AngularJS maintenance and Next.js feature delivery on one platform roadmap.

Technical wiki

Documented decisions, patterns, and workflows so context outlives any single contributor.

Reflection

What this journey reinforced

Lead Developer at UMGC has been less about hero features and more about sequencing products that compound.

Modernization is a relay, not a rewrite

Keeping AngularJS stable while Next.js grows feature-by-feature taught me that parallel paths beat big-bang cutovers when production cannot pause.

Pipelines are products too

Branch strategy and CI/CD are invisible until they break. Treating publish reliability — above 88% success — as a deliverable changed how the team plans releases.

Agents need orchestration, not just prompts

The copilot program worked because each team got a starter kit, a CLI, and Azure DevOps hooks — not because someone pasted a system message into a chat window.

Outcomes

Delivery reliability

Publish pipeline health became a tracked outcome once CI/CD gates and review culture landed.

2025

88%+ publish success · low failure rate