返回首页
最新
I have a decade of mixed experience: 10+ years in IT, 2 years on a professional motorsports team (Tony Kart, Ducati, Miata), and 4 years in software (refactoring old .NET to Spring Boot/Angular and some greenfield projects). I self-taught my way into dev and finished a CS degree while working.<p>I’m currently at a {big_slow_corp} and feeling the AI squeeze trying to switch jobs. My exp is not public facing so the “experience building scalable apps to disrupt the market” requirement I think is my weakness. I’ve also realized my heart isn’t in web frameworks. I’m the guy bored on a plane talking to his gf about flap software, throttle response, and suspension geometry like a kid showing a toy to his mom.<p>I want to move "closer to the metal", ECUs, controlling machinery, I’d love ML applied to machinery, even infotainments!, but my professional resume is strictly high-level. My tinkering experience includes building MegaSquirts and using ESP32s for signal filtering and logic, data loggers and wiring things you’d find in a race car, but I currently lack a project car to demonstrate new work.<p>How can I bridge the gap from Java/Angular to automotive software?<p>Does a decade of IT and 4 in backend experience carry weight in the "Software Defined Vehicle" world, or am I starting at zero?<p>What "proof of work" can I build to wow recruiters at places like Toyota Research or Tesla (any brand with US presence really) without a physical car to hack on?<p>I'm ready to pivot to where my passion actually is. Any advice is appreciated.
I built PurchaseKit to add in-app purchases to Hotwire Native apps without writing native code.<p>Adding subscriptions to a native app means learning StoreKit and Play Billing. Two platforms, two languages, two sets of documentation. Then you try to wire it all up to your Rails backend.<p>PurchaseKit gives you drop-in bridge components for iOS and Android. Your paywall stays in ERB. No Swift, no Kotlin.<p>It also normalizes Apple and Google webhooks into four events (created, updated, canceled, expired) so you don't have to deal with their different formats.<p>Works with the Pay gem or your own models.<p>Free for up to 10 paying customers, $99/mo after.<p><a href="https://purchasekit.dev" rel="nofollow">https://purchasekit.dev</a>
Scan your website and get insights for better SEO and AI visibility.
PokerBench is my attempt at a new LLM benchmark wherein frontier models play Texas Hold'em in an arena setting. It also features a simulator to view individual games and observe how the different models reason about poker strategy. Opus/Haiku, Gemini Pro/Flash, GPT-5.2/5 mini, and Grok 4.1 Fast Reasoning have all been included.<p>All code -> <a href="https://github.com/JoeAzar/pokerbench" rel="nofollow">https://github.com/JoeAzar/pokerbench</a>