Erick Castillo

Challenge

The ELA platform sat at the intersection of three groups that didn't naturally speak the same language: the federal client (FEMA), Corpay business stakeholders, and cross-functional IT groups including infrastructure, security, and DBA teams. Requirements arrived in business terms, infrastructure changes arrived in ticket-speak, and the development team needed both translated into actionable engineering work without losing fidelity. Misalignment on a federal contract — a misread requirement, a missed compliance constraint, a surprise infrastructure dependency — would have meant rework, blown milestones, or worse. Someone had to own the translation layer so the team could ship.

Solution

Served as the primary liaison between Corpay business stakeholders, FEMA-facing contacts, and cross-functional IT groups for the full ELA engagement. Coordinated meetings, gathered requirements, surfaced technical constraints early, and translated between business language and engineering language in both directions. Made sure the development team had clear, unambiguous work to do and made sure stakeholders had honest, accurate status — including when something was at risk.

Impact

Kept the engagement aligned across three groups that had different vocabularies, different incentives, and different definitions of "done." The development team got clear, unambiguous work because requirements were translated and validated before they hit the backlog. Stakeholders got honest status, including when something was at risk, which built trust faster than optimistic reporting would have. Cross-functional dependencies (infrastructure, DBA, security) were surfaced early enough to schedule around rather than absorb as surprises. The documentation trail — meeting scripts, ADRs, Jira tickets — gave the engagement a defensible record on a federal contract where audit-readiness matters.

Technologies

Challenge

Hotel image updates on the portal depend entirely on a manual process that requires support staff to access production servers directly and modify database records by hand. This creates a bottleneck where routine image changes are delayed behind ticket queues, introduces risk of human error during manual data entry, and ties up support resources on tasks that could be handled by administrators themselves.

Solution

Introduce a built-in image management capability within the existing admin portal that enables authorized users to upload, preview, and save hotel images through a web-based interface. The system handles file storage and database updates automatically, eliminating the need for direct server access or manual database modifications.

Impact

 Administrators can now manage hotel images independently without submitting support tickets or waiting for manual intervention. The change is self-contained within the admin module, with no modifications to existing business logic, database schema, or the public-facing site. Support teams are freed from routine image update requests, and turnaround time for hotel image changes is reduced from days to minutes.

Technologies

Challenge

The FEMA ELA platform's primary data store ran on MySQL 5.7, which was approaching end-of-life and lacked the security, performance, and JSON/CTE features the .NET application increasingly relied on. On a federal platform supporting emergency lodging during disaster events, the database also had to demonstrate verified high-availability failover — a documented, repeatable test, not a hope. Upgrading in place risked breaking EF Core data access, stored procedures, and existing query plans, and a botched failover during an active disaster response would have been a serious incident.

Solution

Owned the MySQL 5.7 → 8.0 upgrade plan end-to-end and ran the HA failover validation exercise alongside it. Authored the TDD covering compatibility analysis, upgrade path, rollback strategy, and the failover test protocol. Coordinated with Corpay infrastructure, the FEMA stakeholders, and the dev team to schedule the cutover, then executed and signed off on the failover drill against the upgraded cluster.

Impact

Brought the ELA platform onto a supported MySQL version with stronger security defaults and modern SQL features, eliminating EOL risk on a federal contract. The HA failover drill produced a signed-off, repeatable runbook — meaning the on-call team had verified evidence of failover behavior rather than a theoretical design. EF Core data access continued working without regressions across the full integration test suite, and the Azure Blob backup pipeline gave the platform a clean DR posture going into disaster season.

Technologies

Challenge

The five-person ELA development team was shipping .NET features at a traditional pace — handwritten CQRS handlers, manually scaffolded EF Core configurations, and inconsistent test coverage across modules. AI tools were available to the organization but used inconsistently: some developers leaned on Copilot for autocomplete, others ignored it entirely, and no one had a shared convention for prompts, context, or AI-assisted code review. The result was uneven velocity, duplicated effort across the team, and no defensible way to measure whether AI tooling was actually helping. On a federal contract with hard delivery milestones, the team needed a repeatable workflow rather than ad-hoc AI use.

Solution

Defined and rolled out team-wide AI-native engineering standards covering spec-first development, prompt and context conventions, and AI-augmented code review checkpoints. Integrated Claude Code, GitHub Copilot, and M365 Copilot into the daily workflow with clear guardrails on what each tool was best for. Established markdown-based Technical Design Documents as the single source of truth that AI agents would consume to scaffold new modules, then trained the team on the workflow through mob-coding sessions and code reviews.

Impact

Shifted the team from ad-hoc AI use to a defined, repeatable workflow with shared conventions and review gates. Time-to-first-working-build for new modules dropped meaningfully once TDD-driven scaffolding became the default — handlers, EF Core configs, and tests generated from spec instead of typed by hand. Code review quality improved because reviewers had a clear checklist for AI-generated code, and onboarding got faster because the standards doc and CLAUDE.md context files gave new developers a runway. The team built a defensible story for leadership that AI tooling wasn't just autocomplete — it was a productivity multiplier with measurable gates.

Technologies

Challenge

The ELA platform's five-developer team came in with mixed experience levels and uneven exposure to the codebase, Clean Architecture conventions, and the modern .NET stack the platform was moving toward. On a federal contract with hard milestones, the team couldn't afford a slow, sink-or-swim onboarding curve — knowledge had to transfer fast, and it had to stick. Beyond onboarding, there was no shared rhythm for code review, no consistent forum for surfacing tricky architectural decisions, and no defensible record that mentorship was actually happening. Skills gaps left unaddressed would have shown up later as production defects on a 24/7 government-facing platform.

Solution

Built and ran a structured knowledge-transfer and mentorship program for the five-developer team covering KT sessions, mob-coding exercises, and code reviews as deliberate teaching moments rather than just gates. Made skill development a tracked part of the engagement rather than a side effect, and used the program to enforce Clean Architecture, CQRS, and the team's AI-native engineering standards consistently across the team.

Impact

Built a team that could ship Clean Architecture .NET features at a consistent quality bar rather than a per-developer one. KT sessions and mob-coding gave the team a shared mental model of the ELA platform, which cut down on architecture-drift comments in code review and made onboarding new tickets faster. Code review shifted from gatekeeping to teaching, which raised the floor on code quality without slowing throughput. The rotation of session leadership built bench depth — more than one person could explain any given subsystem — which mattered on a federal platform where bus-factor risk was real. Mentorship deliverables (AI-native workflow, spec-first practice) gave team members skills that ported beyond the engagement.

Technologies

Challenge

The ELA platform's REST API had endpoints that performed acceptably under normal traffic but degraded under the spikes the platform sees during active disaster-response windows. Common issues: N+1 query patterns in EF Core, list endpoints returning full entities with every related collection eagerly loaded, async methods that quietly ran synchronously because of a single blocking call, and pagination implemented with Skip/Take against large tables where offset-based paging gets progressively slower. On a federal platform where a traffic spike correlates with the worst possible moment to slow down, the API needed to be tightened before the next event, not after.

Solution

Profiled and hardened ASP.NET Core REST endpoints across the ELA platform, fixing N+1 query patterns, replacing offset pagination with cursor-based pagination on hot list endpoints, enforcing async-all-the-way-down, and adding response compression and output caching where it was safe. Used MiniProfiler and EF Core query logging to surface the worst offenders, fixed them at the handler and EF Core configuration level rather than papering over with caching, then added caching on top of an already-fast baseline.

Impact

Hot endpoints stopped degrading under load because the fixes happened at the query and async layers rather than being hidden behind caching. Cursor-based pagination meant deep pages performed as well as shallow ones, which mattered for admin and reporting endpoints that legitimately needed to scroll far into result sets. Async-all-the-way-down freed up request threads under spike traffic, which raised the platform's effective throughput on the same infrastructure. Response compression and output caching layered on top gave additional headroom without masking underlying problems. Structured logging plus Azure Monitor meant the next regression would be easy to find — observability became a real capability, not a checkbox.

Technologies

Challenge

The current system relies on unsupported and end-of-life technologies, which introduces significant risk and operational limitations:<div><br></div><div>PHP 7.2 is end-of-life, meaning it no longer receives security updates or bug fixes, exposing the application to known and future vulnerabilities.</div><div><br></div><div>MySQL 5.7 is no longer supported, increasing the risk of data integrity issues, performance degradation, and security exploits.</div><div><br></div><div>Known vulnerabilities remain unpatched, leaving the system susceptible to cyberattacks, compliance violations, and potential service disruptions.</div><div><br></div><div>These issues collectively increase security risk, maintenance costs, and technical debt, while limiting the ability to scale or integrate modern features.</div>

Solution

Modernize the technology stack by upgrading to supported, secure, and actively maintained versions:<div><br></div><div>Upgrade PHP to a currently supported LTS version (e.g., PHP 8.1 or 8.2) to ensure ongoing security patches and performance improvements.</div><div><br></div><div>Migrate MySQL 5.7 to MySQL 8.0 (or an alternative supported database such as PostgreSQL) to gain enhanced security, improved query performance, and long-term support.</div><div><br></div><div>Implement a vulnerability management and patching strategy to ensure all dependencies and system components remain up to date.</div><div><br></div><div>This modernization will reduce security risks, improve system performance, and ensure long-term maintainability.</div>

Impact

After implementation, the organization will achieve:<div><br></div><div>Significantly reduced security risk through supported software and active patching.</div><div><br></div><div>Improved system stability and performance, with faster execution and optimized database operations.</div><div><br></div><div>Regained compliance readiness, supporting audits and regulatory requirements.</div><div><br></div><div>Lower long-term maintenance costs due to reduced technical debt and easier upgrades.</div><div><br></div><div>Future-ready architecture, enabling scalability, new features, and integrations.</div>

Technologies

Challenge

The existing monolithic .NET Framework 4.8 application struggled to scale under growing user demand (10,000+ daily users), suffered from tightly coupled components that slowed feature delivery, and lacked proper API governance. Issue detection was slow and reactive, leading to extended downtime and poor user experience.

Solution

Redesigned the application architecture from monolithic to microservices using Docker and Azure Kubernetes Service (AKS). Migrated legacy .NET Framework 4.8 applications to .NET Core and integrated Java microservices. Implemented Azure API Management for centralized API governance and Azure Monitor with Application Insights for proactive performance monitoring. Built serverless event-driven applications using Azure Functions.

Impact

Improved system scalability by 40% and enabled independent service deployments. Boosted application performance by 30% through the .NET Core migration. Reduced issue detection time by 50% and maintained 99.9% application uptime. Reduced infrastructure overhead through serverless architecture. Improved API governance and security across the platform.

Technologies

Challenge

Mission-critical retail operations spanning 200+ stores required high-availability systems capable of processing 100K+ daily transactions. The existing infrastructure had high cloud costs, lacked Infrastructure as Code practices, and suffered from data latency issues across distributed environments. Junior developers needed upskilling on cloud architecture patterns.

Solution

Developed cloud-native applications using Azure Functions, Service Bus, and Event Grid for reliable message-driven architecture. Integrated CosmosDB for high-throughput, geo-redundant data operations. Implemented CI/CD pipelines with Infrastructure as Code (ARM templates). Executed Azure cost optimization strategies and built reporting solutions with Power BI. Mentored junior developers on cloud best practices.

Impact

Achieved 25% reduction in Azure infrastructure costs through optimization strategies. Ensured high availability and performance at scale for 200+ retail stores. Enabled low-latency data access across distributed environments with CosmosDB. Reduced manual release overhead through automated deployments and Infrastructure as Code. Accelerated team capability growth through mentorship on cloud architecture.

Technologies

Challenge

The organization managed 100+ applications with inconsistent testing practices, low code coverage, and high bug escape rates. Bug detection was slow, quality assurance was largely manual, and application performance suffered from unoptimized code paths, resulting in slow response times on critical endpoints.

Solution

Developed a comprehensive unit testing framework using MSTest and NUnit to standardize testing across the enterprise. Implemented automated testing strategies integrated with CI/CD pipelines. Performed code optimization and profiling to improve application performance across critical endpoints.

Impact

Achieved 85%+ code coverage across the enterprise application portfolio. Reduced bug detection time by 60%, significantly improving software quality. Improved response times by 35% on critical application endpoints. Accelerated release cycles through automated testing and reliable deployment processes.

Technologies

Challenge

Financial institutions and enterprises needed compliant lease accounting solutions to meet GAAP standards. Manual data processing between systems was time-consuming and error-prone, and the existing application lacked real-time data capabilities and code reusability, hindering development velocity.

Solution

Developed custom financial reporting solutions using ASP.NET Core with React and Angular frontends for lease accounting automation. Built complex SQL queries and stored procedures for compliance reporting. Integrated third-party APIs for automated data synchronization. Implemented component-based architecture with RESTful APIs and Vue.js for real-time data updates

Impact

Delivered audit-ready financial compliance reporting supporting GAAP standards. Reduced manual data processing time by 50% through automated third-party API integrations. Improved code reusability and scalability through component-based architecture. Enabled real-time data updates for faster decision-making.

Technologies