2adays.com

I developed the Rails application for 2adays.com, a website that helps student athletes through the college recruiting process. My work involved replacing an existing coach and school search directory with a fully custom one. The Rails application is responsible for pulling information from a third party API for coaches and schools, and formatting the date for use in student searches and ratings. Recently, I implemented a new recurring subscription service.

Background

2adays.com is a website that helps student athletes through the college recruiting process. Current college-level athletes are able to rate their coaches through a survey-based system. These surveys are tabulated and ratings are generated for each coach. High school athletes and their parents are able to search these coaches and surveys so that they can make better choices as they enter the college athletic world.

2aDays approached me in late 2016 to build a replacement for their directory service using Rails. This application would be handed off to another developer for long term maintenance. Best practices were a deliverable.

This application best represents the work I do. My contact at 2aDays gave me loose guidelines and expectations and left me to do my best work. A little of everything I do with Rails went into this one:

• Front end design with Bootstrap, SCSS, a splash of CoffeeScript, and good old .html.erb.
• A healthy dose of ActiveRecord.
• I18n. Totally worth it, even for just one language.
• Mirroring data from an (often quirky) “REST” API.
• Thorough testing. But not too much, you know?
• Server setup with Vagrant.
• Documentation. I made sure the next developer would know how to get around.
• I was able to ship the app in 6 weeks working nights and weekends. The 2aDays team was very happy with the final result, and the developer that took over maintenance has done so with ease.

Technical Overview

Architecture

The app is a coach and school rating platform for 2adays.com. Athletes rate college coaches through SurveyGizmo surveys, and the results are displayed alongside coach/school profiles imported from the Collegiate Directories API.

Data pipeline:

1. Rake tasks import sports, coaches, schools, conferences, and divisions from the Collegiate Directories API using a base importer class with a template pattern. Each import logs structured status (skip/update/create) and runs transactionally.
2. SurveyGizmo posts completed rating surveys to a webhook endpoint (POST /ratings). Each rating includes 7 trait scores (Accessible, Cool, Direct, Honest, Intense, Knowledgable, Motivational) on a 0-5 scale plus free-text comments, stored as serialized JSON.
3. A blog feed is scraped from a WordPress site via Nokogiri and cached hourly.

Key models and relationships:

• Coach → has many Profile records (one per sport/school combination)
• Profile → belongs to Coach, School, Sport, Conference, Division; has many Rating records
• Rating → stores serialized JSON data with trait scores and comments; calculates an overall score on save via before_validation callback
• Profile uses a status state machine (enabled → pending → disabled) to handle coaches that appear/disappear between API imports

Search: Custom SQL implementation using UNION ALL across coaches and schools tables, with MySQL SOUNDS LIKE for phonetic matching and LIKE for fuzzy matching. Pagination handled with Kaminari against raw connection.select_all results. An “I’m Feeling Lucky” feature auto-redirects when a search returns a single result.

Caching strategy: Redis-backed in production with layered TTLs — API responses cached indefinitely, blog feed cached for 1 hour, view fragments cached for coach profiles and the “Best Coaches” homepage widget (top 5 coaches with a minimum of 5 ratings).

Avatar system: Paperclip with model-specific default images, multiple sizes (64x64 thumb, 300x300 medium), local filesystem storage in development and S3 in production.

I18n: Full internationalization support for a single language — used for sport names, profile titles (“Men’s/Women’s/Coed {Sport} Coach”), and UI strings. Worth it for consistent string management and easy future localization.

Deployment

12-factor app configuration via dotenv. Environment variables for database (DATABASE_URL), Redis (REDIS_URI), Collegiate Directories API token, S3 credentials, and SurveyGizmo survey link template. Logs to STDOUT. Scheduled imports and blog scraping via cron. Vagrant provisioning scripts and thorough documentation ensured a smooth handoff to the maintenance developer.