I’m 36. From Kyiv.
Right now, my city is suffocating under a heatwave. At night, the air raid sirens wail, and the psychological weight of a war with no end in sight presses down on every decision. I am typing this on an AMD A4-4020 with 8GB of RAM — a machine so weak it chokes on modern web browsers.
I am not a traditional developer. I am a vibe coder. To me, vibe coding isn't just a buzzword; it's a philosophy. It’s the ability to translate a feeling, an idea, or a complex vision into a precise prompt for a machine. It’s about orchestrating AI to bring a concept to life, rather than getting bogged down in the syntax of writing code from scratch. By day, I work as a webmaster and system architect, navigating a full-time job and a complex medical regimen.
Yet, I am entering the RevenueCat Shipaton 2026 hackathon. Alone. Against teams with M2 Macs and budgets.
Why? Because I don't just want to build an app. I want to prove that constraints don't kill projects. Fragility does.
This is the story of NICHLYST — a slow narrative survival-archive sim for Android — and the 128-step antifragile plan I wrote to ensure it actually ships by September 28th.
The Concept: Survival as a Gameplay Loop
In NICHLYST, you are an archivist in a besieged city. You don't fight. You record. Every day, someone knocks on your door with a story. You must choose: record the full truth and risk your safety, or edit it to survive. The consequences don't arrive immediately — they arrive days later, via a real-time push notification, forcing you to live with the anxiety of your choices.
It’s a game about documentation, truth, and the civilian experience of conflict.
Phase 0: The Pre-Mortem
Most hackathon projects die because developers start coding on day one. I started by writing a 128-step plan, but before writing a single line of code, I forced myself to do a pre-mortem.
I listed the 10 most likely ways NICHLYST would fail. Here they are, raw and unfiltered:
- My weak local PC freezing during an Android build.
- Solo development bottlenecks.
- Missed evenings due to fatigue or schedule conflicts.
- Google Play App Review delays.
- AI art style mismatch or inconsistency.
- RevenueCat sandbox failure or SDK integration issues.
- Missing the hard deadline entirely due to scope creep.
- Psychological burnout.
- Catastrophic loss of content or code.
- Conflict with an existing 32-day evening schedule I must maintain.
Then, I applied Nassim Taleb's Antifragility framework. I didn't just want to mitigate these risks; I wanted the project to benefit from the stress.
The Barbell Strategy: I allocated 90% of my time to the core, guaranteed MVP. The other 10% is strictly time-boxed for bonus categories. The "middle ground" — medium-risk, medium-reward tasks — is entirely eliminated.
Stop-Loss Dates: I hardcoded triggers into my plan. If the AI art pipeline isn't done by August 26th, I don't argue with myself. I automatically cut the art scope to a stark, 1-bit CSS visual filter and move on. Emotion has no place in the schedule.
The 23:00 Rule and the Reality of Mental Health
My plan includes a strict rule: work ceases at 23:00. A tired solo dev creates more bugs than they fix. Sleep is a strategic investment.
But that’s the theory. The reality is messier.
Sometimes external factors — late shifts at my day job, or the air defense activity in Kyiv — make the 23:00 deadline impossible. And then there are the internal factors. I live with GAD and depression. There are days when there are no missiles, no extra work, but my brain simply refuses to cooperate. The energy tank is empty. On those days, the only achievable victory is making it to the shower and getting into bed.
An antifragile plan can’t be a rigid, brittle clockwork. It must absorb these bad days. If I miss an evening due to a depressive episode, the Stop-Loss dates protect the project. The schedule bends, but it doesn't break. It accommodates the reality of being human.
Phase 1: Infrastructure and the Art of Being Broke
I operate on a strict, near-zero budget. My monthly income is about $700 USD, and rent consumes half of it. The only financial investment I made was paying the $25 Google Play Console registration fee — a serious amount of money for me right now.
The Shipaton sponsors heavily promoted Replit for cloud builds. Replit even offers a 40% discount for hackathon participants. But even with the discount, a Replit Pro subscription is financially impossible for me.
So, I engineered a workaround. I set up a local development environment on my Xubuntu 26.04 LTS machine using VS Code. For the heavy lifting — compiling the Capacitor Android APK — I connected my GitHub repo to Codemagic's free tier.
The setup wasn't painless. Capacitor v7 requires Java 21, but Codemagic defaulted to an older version. My first cloud build failed with error: invalid source release: 21. A quick edit to the codemagic.yaml file fixed the environment variables, and the build went green.
When I sideloaded that blank, debug APK onto my Xiaomi Redmi Note 9 Pro and saw it launch... it wasn't just a blank screen. It was proof that a pipeline running from a $0 local machine to a $0 cloud builder to a physical device actually worked. The cycle was closed before I wrote a single line of game logic.
The Google Play 20-Tester Bureaucracy
Then came the infrastructure hurdle. Google Play now mandates a 14-day closed testing track with 20 active testers before a new personal account can publish to production.
This is a bureaucratic nightmare for a solo dev with a ticking clock. If I wait until the game is done to start this timer, I will miss the deadline. So, I built a Google Form and posted it in the Shipaton Discord, my social channels, and sent it to colleagues.
The result? Humbling.
I got 3 emails from colleagues — my most loyal audience. From the Discord community? Zero. After 24 hours, the form remains agonizingly empty.
I try not to succumb to panic. I try to filter out the noise, both from the outside world and my own anxiety. But the math is unforgiving: I need 20 testers, and I have 3.
So, I am putting this directly into this devlog. If you are reading this, have an Android device, and 2 minutes to spare, please help a fellow builder bypass this bureaucratic wall.
👉 https://forms.gle/4hJkWsmuyKp3Tgtr8
You will just need to accept an invite from Google Play later, download the app, and keep it installed for 14 days. No deep QA required. Just your presence to satisfy the algorithm.
The Unseen Labor
Right now, NICHLYST is just a 128-step plan, a handful of .md files tracking risks and deadlines, a GitHub repository, and a codemagic.yaml configuration. No art, no story code, no RevenueCat paywalls yet.
But the foundation is poured. It is not fragile.
The day before yesterday, after a 10-hour shift at my day job and helping my wife with her transcription work, I had zero energy left. I didn't code. I didn't force it. I followed the rule: I updated my PLAN_STATE.md log with one sentence, did a git commit, and went to sleep.
In a hackathon, survival isn't about sprinting. It's about building a system that allows you to keep walking, even when the ground is shaking.
NICHLYST is coming. And it will ship on time.
Follow this devlog for the #BuildInPublic journey.
Top comments (0)