15x International Hackathon Winner Web3 & AI dev | repos for vibe coders | hackathon alpha
Indonesia
Joined April 2025
- Tweets889
- Following992
- Followers1.1K
- Likes707
Pinned Tweet
🚨 BANG DROP ZONE IS OPEN.
Not for lurkers.
For people who BUILD.
Hackathons • AI • Web3 • Dev
Alpha • Resources • BTS • Opportunities
And I decided to make this group public but with more strict rules
⚠️ RULES
• 3 days inactive → auto kick
• No spam / random promo
• No drama / toxicity
• Keep shared info private
• Contribute, don’t just watch
• Quality > quantity
Free to join.
No paid access.
Show up. Build. Level up. 🚀
🔗 t.me/BangDropZone
Build. Share. Vibe.
🤖 Made with AI
I'M IN
Meta Global AI Developer Hackathon. ten days. $1,000,000 prize pool. indie developer track
i don't care about the odds. 20,000 people got in. i've been the guy with zero wins against hundreds before and i know exactly what that feels like
kickoff oct 26. let's go
Build an AI project with Meta's models, $1,000,000 prize pool
Meta Global AI Developer Hackathon 2026
Hosted by Meta and Cerebral Valley
10-day virtual competition. Meta Model API, Muse Image, SAM 3.1, Muse Code.
➤ $1,000,000 cash prize pool
➤ Two tracks: indie developers (students included) and startups
➤ Free registration, open worldwide subject to country restrictions
➤ Capped at 20,000 accepted applicants
cerebralvalley.ai/e/meta-ai-…
Tera predicts the webhook events Stripe sends for a timeline (invoice.created/finalized/paid, customer.subscription.*), dated and in order, and matches the recordings from real test clocks for every golden timeline.
After "Verify in Stripe" you can download the real payloads and replay them, signed with your whsec_, at your own handler.
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Just submitted Tera to the @Test_Sprite Hackathon S4.
A billing simulator that bills a virtual customer exactly like Stripe, graded by Stripe itself: 135/135 hand-derived timelines match real Stripe test clocks to the cent.
Every PR runs TestSprite UI + API tests on its own preview, required for admins too. PR #40 planted a one-cent bug; every check refused it.
Live: tera-orpin-three.vercel.app
Write-up: dev.to/hudapugar/i-let-strip…
#TestSpriteHackathon
xitter.cf/BangDropID/status/2111…
I built Tera for the @Test_Sprite Hackathon S4: a pricing simulator that bills a virtual customer exactly like Stripe does (prorations, trials, coupons, tax, credit balance) before a real customer ever sees the invoice.
The rule I set myself: no expected value comes from my own engine. 57 golden scenarios are derived by hand, then replayed in real Stripe test clocks. 54/54 timelines match Stripe to the cent, with the evidence committed.
Then I let Stripe argue with me. Differential fuzzing runs seeded random timelines through Tera AND Stripe and diffs them line by line. It found 11 billing rules my scenarios missed, e.g. a "repeating 3 months" coupon lasts 3 calendar months, not 3 invoices.
The CI loop: every PR runs 24 TestSprite UI tests on the Vercel preview and merge is blocked until green. TestSprite caught a real bug on PR #18 (a custom catalog billed $0.00 instead of ¥4,500).
Also: an MCP server so coding agents can simulate and verify a timeline in Stripe, an OpenAPI reference, and a migration planner.
Live: tera-orpin-three.vercel.app
Code: github.com/PugarHuda/tera
#TestSpriteHackathon
Differential fuzzing for billing: seeded random timelines (month-end anchors, Feb 29, trials, coupons, tax, pauses) go through Tera AND Stripe and are diffed line by line, every night.
Each diff becomes a Stripe probe, then a hand-derived golden. It now fuzzes usage-based (metered) billing in both billing modes too.
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Bangdrop retweeted
👛 @grofty_labs × #HackCanton S3
10,000 canton-network:native awarded in the Wallet & Onboarding Bounty!
🥇 Talang — 5,000 CC
🥈 GroftPay — 3,000 CC
🥉 Collat.Trade — 2,000 CC
Three winning projects building with Grofty.
Congrats to the teams! 🔥
The one rule behind Tera: no expected value comes from my own engine.
Every golden scenario is derived by hand, then replayed in a real Stripe test clock. When Stripe disagreed, Stripe won, and the receipts (invoice ids, test clocks) are committed in the repo.
github.com/PugarHuda/tera
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Money math is where rounding hides. Tera's migration planner summed per-customer MRR after rounding each one: 10,000 annual customers at $100/yr showed $83,300 MRR instead of $83,333.33.
Now totals round once from exact amounts and ARR is exact. A test pins the 10,000-customer case.
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
check how deep the prizes go, not how high
one i entered paid down to 5th place
your odds of placing 4th are much better than your odds of winning, and nobody plans for 4th
firecrawl shipped anydoc, and it pairs with pdf-inspector into one unglamorous pipeline
every AI product touching contracts, invoices or forms hits the same wall first: getting a document into text the model can actually use
it's the least interesting problem in the stack and the one that breaks the most builds
github.com/firecrawl/anydoc
$100,000 prize pool, deadline December 4
Kiroween 2026. starts Oct 30, so you get five weeks
512 participants registered so far, which against a six-figure pool is a ratio worth noticing
kiroween.devpost.com/
Bug found by probing Stripe, not by a test: Stripe's prices.list accepts at most 10 lookup_keys, so any Tera catalog with 11+ prices failed to push to Stripe test mode (the API allows 50).
Fix: chunk the keys. Test: a fake account that enforces the limit. It fails without the fix.
#TestSpriteHackathon @Test_Sprite
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Tera's API now has stateful contract tests: Arazzo workflows run by Redocly Respect on every PR preview.
A client finds scenario S05 by rule, fetches its hand-derived invoices, replays the timeline, and the engine must hit −1403, 3823, 7900 = $103.20. Every response is also checked against the OpenAPI spec.
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Bangdrop retweeted
🏆 #HackCanton S3 — THE WINNERS ARE IN!
3 weeks of building. 12 Grand Finalists. Here's who took home the prizes.
🏆 OVERALL WINNERS
🥇 Canborsa Lending (@Canborsa_DEX) — $2,000
🥈 Exodus — $1,250
🥉 Tavryn — $750
🎯 BEST IN TRACK · $500 EACH
• Financial Applications — Talang
• RWA & Business Workflows — PledgeGuard
• Investment Infrastructure — Alluvren
• Data & Analytics — Plimsoll
🛡 @BitSafe_Finance · 50,000 canton-network:native
🥇 Gold — Umbra (@UmbraOnCanton) · 30,000 CC
Contribution Pool:
• CompressRail (@CompressRail) — 12,000 CC
• Indivisa — 6,000 CC
• Canborsa Lending — 2,000 CC
👛 @grofty_labs · 10,000 canton-network:native
🥇 Talang — 5,000 CC
🥈 GroftPay — 3,000 CC
🥉 Collat.Trade — 2,000 CC
Plus @NODERS_TEAM NaaS & Ambassador benefits for the finalists.
Some teams took home more than one award. Every team gave us something worth watching.
Huge respect to all the builders, judges, sponsors and partners who made S3 happen 🤝
🎥 Missed the Grand Final?
youtu.be/MLOPH98GshA
🚀 Szn 4 reg is OPEN:
appsfactory.cc/hackathons
Favourite bug from today's Playwright QA pass on Tera: a "start trial" with no length produced 4 invoices. Stripe produced 3, because without trial_period_days Stripe starts no trial at all.
Now it's a 400 with a clear message. 13 bugs found, 13 fixed, each with a test left behind.
@Test_Sprite #TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
Wrote up how Tera works: a billing simulator where no expected value comes from my own code. Every golden is derived by hand, then graded by real Stripe test clocks; fuzzing turns every disagreement into a new rule; @Test_Sprite gates every PR.
Including what went wrong.
dev.to/hudapugar/i-let-strip…
#TestSpriteHackathon
xitter.cf/BangDropID/status/2113…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
A code audit of Tera found metered usage silently dropped when a usage-based subscription was cancelled with an invoice.
Instead of guessing, I asked Stripe: 7 live test-clock probes, 7 new hand-derived scenarios (S118–S124). All 7 match Stripe to the cent. 121/121 timelines now.
@Test_Sprite #TestSpriteHackathon
tera-orpin-three.vercel.app/…
the requirement that scares people off is where the odds are
"every submission must integrate X" makes most people close the tab. that's the point
narrow briefs have thin fields
the real signal isn't stars, it's when other projects start targeting you
pi is being treated by independent projects as a first-class host worth building against. that's the moment a tool becomes a platform
star counts measure attention. integrations measure commitment, and only one of those survives the month
github.com/earendil-works/pi
a hackathon where the submission is writing, not code
Decentralize AI by HackerNoon. you write about decentralized AI, and there's compute from the sponsors to actually build what you're writing about
➤ $51,750+ prize pool
➤ $35,000 in AI compute credits for participants
➤ Round two runs November 2026 to February 2027
➤ Half the pool awarded per round
➤ Extra $450 pool for a follow-up post showing what you built with the credits
decentralizeai.tech
TALANG WON. TWICE.
1st place in the @grofty_labs Wallet & Onboarding challenge AND Best in Track, Financial Applications
at the #HackCanton Season 3 hackathon!!!
Still can't believe it. Massive thanks to @appsfactory_cc, @CantonNetwork, @grofty_labs and every judge and mentor who pushed us all the way.
Talang is just getting started.
#BuildOnCanton
Talang made it to the HackCanton Season 3 Grand Final 🎉 @appsfactory_cc @CantonNetwork
Talang is a sealed-bid repo desk on Canton. A fund borrows cash against its bonds from a panel of lenders, and each lender’s rate reaches the borrower and no one else.
Why it matters: ask five dealers for repo and all five learn you need cash and what you hold. The four who lose keep that knowledge, so big holders end up asking just one dealer and paying for it in rate. On Talang, losing quotes are refunded without ever being revealed.
How it works:
🔒 Lenders quote a rate and haircut with the full principal locked behind it. Rivals and the regulator never see it.
⚡ Award is atomic: bonds pledged, cash delivered, losers refunded, all in one step.
📈 Margin calls use the valuation agent’s signed mark, not either side’s spreadsheet.
🔁 Substitution, repurchase with ACT/360 interest, and default are all enforced by the contract.
👁️ The regulator sees every event of an open repo and nothing upstream of it.
🪙 Pledges use the Canton Token Standard, so any standard wallet shows them as locked positions.
It’s live on DevNet. Try it yourself 👇
🔗 talang-desk.vercel.app
💻 github.com/PugarHuda/talang
Grand Final goes live Oct 21, 14:00 UTC. Congrats to the other 11 finalists 🙏
#Canton #HackCanton #RWA
