Experience.

next.js · fastapi · postgresql/pgvector · aws ecs · terraform · docker

An internal OptimalDevs tool that turns a repository's own history into something a new developer can actually read, instead of a wall of unfamiliar files.

• Turned GitHub commits, pull requests, and source files into searchable timelines, guided repository tours, repository-aware chat, and plain-language explanations.
• Implemented hybrid retrieval with embeddings, vector search, and PostgreSQL full-text search so answers stay grounded in real repository context.
• Added AI usage controls to keep inference costs bounded per repository.
• Provisioned AWS infrastructure with Terraform, ECS Fargate, RDS PostgreSQL, ECR, ALB, IAM, and CloudWatch.
• Automated releases with Docker images and GitHub OIDC, including private networking, rollback, and cost tracking.

What I learned: retrieval-augmented generation, vector search, infrastructure as code, container orchestration, and cost-aware AI system design.

astro · react · typescript · tailwind css · aws ec2 · caddy · github actions

Client work through OptimalDevs for a Denver home services business. The site is running on a preview deployment; the launch on the client's production domain is still ahead.

• Built service pages, location landing pages, and an online quote funnel backed by Web3Forms.
• Added LocalBusiness, Service, and FAQ JSON-LD, canonical URLs, sitemap generation, and Open Graph metadata for local search.
• Deployed to AWS EC2 with Route 53 DNS, Caddy-managed HTTPS, and GitHub Actions builds for client-ready preview releases.

What I learned: local SEO structure, static site architecture with Astro, structured data, and building against a real client's review cycle.

python · aws sagemaker · glue · athena · xgboost · cdk

A personal ML research project studying alleged coordinated microcap promotion events, using labels drawn from public federal court documents.

• Developing an ML pipeline that flags alleged coordinated promotion events from legally documented labels, market data, and social text features.
• Structuring point-in-time feature engineering to prevent lookahead bias in a time series setting.
• Designing training and evaluation around SageMaker, XGBoost, purged cross-validation, PR-AUC, precision@k, and SHAP.
• Planning reproducible MLOps with Model Monitor, CloudWatch, AWS CDK, and serverless inference.

What I learned: time series validation, imbalanced classification metrics, model explainability, and MLOps practices on AWS.

react · javascript · vite · tailwind css · aws · node.js · github actions

OptimalDevs is the production website for the development collective I founded. The site explains our services, team, partnership model, and contact workflow for small business owners.

• Built a responsive single-page application using React, JavaScript, Vite, Tailwind CSS, and React Router.
• Designed reusable sections for services, team profiles, partnership information, portfolio cards, and contact flow.
• Deployed the frontend using Amazon S3, CloudFront, Route 53, ACM, and Origin Access Control.
• Built a serverless contact form using Node.js, AWS Lambda, API Gateway, Amazon SES, and IAM.
• Implemented deployment automation with GitHub Actions, including lint/build checks, S3 sync, cache headers, and CloudFront invalidation.
• Added production-focused infrastructure including HTTPS, CDN caching, private S3 origin access, and low-cost hosting.

What I learned: frontend architecture, AWS static hosting, serverless backend design, cloud permissions, CI/CD, DNS, SSL/TLS, and production deployment.

go · sqlite · aws ec2 · caddy · github actions · linux

This is my personal portfolio and technical blog platform. The landing site introduces my background and projects, while the blog backend supports dynamic technical posts.

• Built a production Go backend for serving blog posts through a lightweight HTTP server.
• Used SQLite for persistent storage, including blog posts, slugs, summaries, draft state, and publish status.
• Implemented a Markdown publishing pipeline that converts technical posts into safe, syntax-highlighted HTML using Go templates and embedded assets.
• Built admin workflows for creating, editing, publishing, and summarizing posts.
• Deployed the backend on an AWS EC2 Linux server behind Caddy HTTPS.
• Secured the server using firewall rules, key-only SSH, fail2ban, automated security updates, systemd service management, and Caddy reverse proxying.
• Automated deployment with GitHub Actions using cross-compiled Go binaries, service restarts, health checks, and safer release patterns.

What I learned: backend development in Go, Linux server administration, SQLite persistence, reverse proxying, HTTPS, CI/CD, AWS EC2 deployment, and production hardening.

java · spring boot · postgresql · docker · swagger · heroku

A team-based backend project for a distributed multi-user dungeon system. I worked on the player service, which exposed REST endpoints for player-related data.

• Worked within a 7-member Agile team building a distributed MUD system across multiple services.
• Owned the Java Spring Boot player service for accounts, characters, and inventory.
• Built RESTful endpoints backed by PostgreSQL, Spring Data JPA, Hibernate, and Maven.
• Used Swagger UI to document API contracts for partner teams.
• Tested API behavior with MockMvc, JUnit, and Mockito.
• Used Docker-based development scripts and Heroku deployment steps to standardize development and deployment workflow.

What I learned: REST API design, Java backend development, Spring Boot, relational databases, team coordination, service contracts, API testing, and deployment workflows.