BlindspotLab

“Football belongs to the fans.” After Ronaldo’s 2008 Ballon d’Or topped the fan vote, I built THE GOLDEN RUN ⚽🏆 a free football game inspired by his road to 1,000 goals. Score. Compete. Climb the rankings. Chase 1,000. Play: goldenrun.fun

Ronaldo or you? Who's going to hit the first 1000 goals? goldenrun.fun

BlindspotLab get a new home, and i want to share with you everything i worked on layer by layer. the first is the Hero, redesigned with new BlindspotLab wordmark, new H1, six ambient icons, and new content for the hero.

i think i can finally write about this. aitax.fun was my first paid client build as a developer. i met the project as a rough idea. no finished architecture. no complete game. just the direction of what the founder wanted to build. my job was to turn it into something that actually worked end to end. the product became a PVP Bitcoin UP/DOWN game on Robinhood Chain. so I built the frontend, smart contracts, the chainlink data streams integration, round automation, settlement, claims, refunds, Burn Accumulator, admin dashboard, health monitoring, indexing, wallet flows, production deployment, all connected together, and working end to end. but getting there was not smooth. one of the hardest moments came after we went live and rounds kept refunding instead of settling. at first, it looked like an oracle or RPC availability problem. it wasn't. i kept digging. the oracle's priceAt(boundary) function was correctly reverting with ReportUnavailable() when the required historical BTC boundary had not been stored yet. that revert was supposed to tell our coordinator: "fetch the authenticated chainlink historical report, submit it, then continue settlement." but our custom RPC transport was swallowing the EVM revert data. instead of seeing the exact ReportUnavailable() signal, the coordinator saw a generic RpcRequestError. so it treated the situation like an RPC failure. the chainlink report retrieval path never ran. and the round eventually reached its refund deadline. we watched multiple rounds refund. that part hurt. this was mainnet. people were using something I built. every failed settlement felt very different from a failing local test. i was tired, at some point, i was genuinely scared that I had built something that wasn't going to work reliably. but I kept tracing it: - revert bytes - RPC responses - oracle boundaries - coordinator state - retry timing - transaction receipts Until the actual failure became obvious: i fixed the transport to preserve EVM revert data. i matched the exact oracle failure condition instead of treating every failure the same. i changed failed settlement retries from the old long delay to a much faster recovery path. i added stage-aware error reporting. i hardened transaction persistence and recovery. then the lifecycle started doing what it was designed to do: entry → lock → oracle → settlement → claim → burn. that experience changed how I think about building production systems. a successful build is not when the UI looks finished, it is not when the contract compiles, it is not even when everything passes locally, it is when the full system survives reality. - networks fail, - providers time out, - transactions get weird, - external services behave differently from mocks, the important part is whether you can understand the system deeply enough to find the real failure instead of patching the symptom. i also built the operational side so the owner doesn't need to live inside Blockscout. contract parameters can be managed from the admin frontend with the owner wallet. settlement is automated. lifecycle functions remain permissionless. the BurnAccumulator is visible to everyone, and once its threshold is reached, anyone can trigger the burn. the contracts remain the source of truth. this started as my first client gig. there were moments I was tired. moments I was scared. moments I wondered whether i had taken on too much. but i learned more from shipping this system than I would have learned from another ten unfinished side projects. i met AITAX as an idea. now it is a working product. and after all those rounds that refused to settle... we're finally UP. 🚀