Software Engineer @ Blue Line Solutions

Kyle
Mitchell

I build full-stack apps with .NET and React, deploy them with Docker and Kubernetes, and manage infrastructure on Azure. Self-taught, AZ-900 certified, and hired as a software engineer nine months from the beginning of my engineering journey.

View My Work Get In Touch Download Resume

<5mo zero to AKS deployment

9mo from journey start to hired

AZ-900 Azure Certified

IaC Terraform-provisioned infra

About

How I
got here.

I started coding on October 7th, 2025. In under five months, I went from writing my first HTML to deploying a containerized full-stack app on Azure Kubernetes Service with a CI/CD pipeline, automated E2E testing, and cloud infrastructure provisioned through Terraform.

Most people take a few years to get here. I got there faster because I was building real things every single day, not following along with tutorials. Everything on this site is something I actually built and shipped.

I like understanding the whole picture. I've worked across every layer of the stack: React components, .NET APIs, PostgreSQL, Docker containers, Kubernetes clusters, Azure cloud resources. I've debugged all of it and I know how the pieces connect.

I passed the AZ-900 Azure Fundamentals certification in April 2026. Three months later I signed on with Blue Line Solutions as a Software Engineer — nine months after I resumed my journey in software engineering.

Oct 7, 2025

Started Coding

First HTML file. Zero prior programming experience.

Nov 2025

Full-Stack Development

React 19 + Vite frontend, ASP.NET Core 7 Web API, PostgreSQL with Entity Framework Core

Dec 2025

Containerization & Testing

Docker multi-container architecture, 6+ Cypress E2E test suites with isolated test database

Jan–Feb 2026

Cloud & DevOps

Azure DevOps CI/CD pipelines, AKS deployment, ACR, NGINX Ingress Controller

Feb 2026

Infrastructure as Code

Terraform for Azure Blob Storage static hosting + Azure DNS Zone.

Mar 2026

Cloud Cost Optimization

Migrated live production app from Azure AKS to Vercel + Render. Maintained full functionality at zero hosting cost.

Apr 2026

AZ-900 Certified

Passed Microsoft Azure Fundamentals exam. Studied with hands-on Terraform, AKS, and Entra ID experience.

Jul 2026

First Engineering Role

Software Engineer at Blue Line Solutions. Nine months from the beginning of the journey to hired.

Skills

The full stack,
end to end.

⚛️

Frontend

React 19 Vite JavaScript (ES6+) HTML5 / CSS3 Responsive Design SPA Architecture Component Architecture

⚙️

Backend

ASP.NET Core 8 C# RESTful API Design Entity Framework Core PostgreSQL Health Check Endpoints Middleware Pipelines

☁️

Cloud & DevOps

Microsoft Azure AZ-900 Certified Terraform (IaC) Azure Kubernetes Service Azure DevOps CI/CD Docker Docker Compose Kubernetes Azure Container Registry NGINX Ingress Azure Blob Storage Azure DNS

🧪

Testing & Tooling

Cypress E2E Test Isolation Patterns Git Azure Repos kubectl Azure CLI PowerShell VS Code

Featured Project

Task App

Production Full-Stack Task Management System

Built from scratch: React frontend, .NET Core REST API, PostgreSQL database. Containerized with Docker, orchestrated with Kubernetes, deployed through a fully automated CI/CD pipeline. Later migrated from Azure AKS to Vercel + Render to keep a live production deployment running at zero cost.

⚡ Live Demo Available

Tech Stack

React 19 + Vite Frontend

ASP.NET Core 8 REST API

PostgreSQL + EF Core Database

Docker + Compose Containers

Kubernetes (AKS) Orchestration

Azure DevOps CI/CD Pipeline

NGINX Ingress Controller Routing

Key Engineering Work

Noticed unusual activity on the live deployment — XSS attempts and request flooding. Hardened the API with input validation, a 30 req/min rate limiter, and a 100 KB request body cap
Added authentication using ASP.NET Core Identity with JWT bearer tokens and Google & GitHub OAuth. Every task is scoped to the authenticated user with server-side ownership enforcement
Migrated from Azure AKS to Vercel (frontend) + Render (API + PostgreSQL) for a zero-cost live deployment
Designed multi-container microservices architecture with Docker Compose
6+ Cypress E2E test suites: create, delete, check, uncheck, persistence, error-handling
Isolated test database environment using a custom Cypress plugin, zero test contamination
Kubernetes liveness and readiness probes for zero-downtime deployments
NGINX Ingress with path-based routing across multiple services
Azure Container Registry with managed identity, no hardcoded credentials
Full CI/CD: code commit → Docker build → ACR push → AKS deploy
Debugged and resolved AKS VM quota limits and subscription namespace issues
Provisioned Azure Blob Storage + DNS Zone via Terraform IaC

Featured Project

Turing the Other Cheek

Real-Time Social Deduction Game With an AI Impostor

A phone-first party game: 3–8 people join a lobby and one player is secretly an AI under a fake human name. Everyone answers the same casual prompts, then tries to catch the impostor. The AI studies each player's real writing over time and blends into the statistical middle of the room, so the more you play, the harder it gets to catch. Includes a reverse mode where the AI guesses who wrote what.

⚡ Play It Live

Tech Stack

React 19 + Vite Phone-first client

ASP.NET Core 8 + SignalR Real-time game server

PostgreSQL (Supabase) Profiles, stats, samples

Gemini / Groq / Cerebras AI failover chain

Docker + GitHub Actions CI, E2E, load test

Vercel + Render Zero-cost hosting

Key Engineering Work

Server-authoritative state machine over a single SignalR hub: lobbies live in memory behind per-lobby locks, a hosted service ticks every timer, client clocks are never trusted
Style profiles: after each game a player's answers feed a background summarizer, and those summaries are injected into the AI's prompt at the next lobby start
Post-processing pipeline that matches the AI's answers to the room's average length, punctuation and typo habits, and strips anything that reads as AI
Game-theory veto rule: accusations don't reveal a result until a 10s veto window passes, and a veto hides the result forever, so vetoing can't leak the AI's identity
Audited anonymity model with a written threat model: no user IDs in rosters, revealed answers keyed by display name and shuffled, timing jitter on AI replies
AI provider failover (Gemini → Groq → Cerebras) with per-provider quota tracking and a circuit breaker, all on free tiers
Free-tier resilience: boots onto an in-memory store if Postgres is paused and restarts itself onto Postgres between games once it's back
229 backend tests on EF InMemory, Playwright end-to-end games on desktop and mobile viewports, and a k6 load test, all in CI on every push
Admin console with analytics, maintenance mode, free-tier usage tracking and user management behind Google-verified admin claims
Procedural chiptune soundtrack built on the Web Audio API with host-synced moods