VoltageBet QA, week one: Cipher took the contract

Cipher Solutions is on a B2B contract as VoltageBet’s QA. I am Kristiyan Radev. I do the testing. This log is the first public note from that engagement: what they bought, what I refused to call green, and why the sportsbook half of the job had to be built instead of staffed.

VoltageBet already had a gold TestRail run of 181 sportsbook cases. Manual runs of that suite are slow, and they are not comparable week to week. Different people pick different markets, wallets, and kickoff times. “It passed testing” and “we know what was tested” are different sentences. That gap is why they hired Cipher.

What landed

Work on the executor started in early July: login, reporting, then a Kafka slice, then live placement. The suite is live on VoltageBet QA. Playwright Test is the runner now. The numbers are not tidy, and I will put them in a later log. This one is the contract.

Cipher’s model on ciphersolutions.dev is the same as the contract: not a CV, not a seat. A coverage plan in writing, testing inside their sprint and their TestRail, and a sign-off that says what passed, what failed, and what the risk is if they ship anyway.

What VoltageBet actually bought

A seat would have meant I join standups and pick cases off the board. VoltageBet keeps the product, the Jira, the TestRail project, and the QA environment. Cipher owns the coverage and the verdict.

In practice that meant:

• Work in their TestRail, not a parallel Cipher suite they would have to reconcile.
• Map the gold run into an executor they can rerun, not a one-off demo.
• Tell them, in writing, which cases are honest on FIFA-300, which need Betby’s own Java harness, and which are product gaps.
• Never publish into the gold run. The human baseline stays readable.

If automation writes into the reference run, VoltageBet loses the only artefact they can still trust when the nightly and the product disagree. That would have destroyed the thing they hired us to make cheaper to execute.

What they did not buy

They did not buy 181 green ticks, a UI click-farm of the sportsbook SPA, or a stub that forges /bet/win so the suite can finish before lunch. Those are the usual ways a vendor looks busy on a sportsbook.

Cipher’s public offer says never a green tick with nothing behind it. On this account that became rules I put in the repo and did not negotiate later:

1. Do not forge provider callbacks. If Node cannot produce half-won or rollback, the case is parked or it goes through Betby’s Java AQA. It is not a skip-as-pass.
2. A missing FIFA-300 board is a fail. Empty feed is what a player would see. Calling it “blocked environment” would have made Cipher look better and VoltageBet less informed.
3. Missing casino pins fail. Missing config is Cipher’s problem until it is in env.
4. Results go to a new TestRail run. The gold run is read-only.

If those rules make the pass rate look worse than a consultancy dashboard, VoltageBet is still getting a verdict they can act on. A 62% nightly with named buckets (RSA drift, missing campaign pin, picker miss, product preference never published) is usable. A 100% nightly that skipped the hard cases is a report.

Where Cipher stops and VoltageBet starts

VoltageBet is the operator. Cipher Solutions is the QA practice on contract. Betby is the sportsbook provider they already integrated. I do not replace Betby or their Backoffice. I execute against both, classify what comes back, and write TestRail comments an operator can read without opening a JSON dump.

Code 280 (operator declined) is usually a partner RSA key that another harness left dirty. That is environment. Cipher heals it when we can, then banners the operator if we cannot. Code 201 after retry (market gone on FIFA-300) is a miss on the section we agreed to test. Cipher fails the case. Mixing those two in one “flake” column would be covering for the stack.

Next logs

This is log 1 of a VoltageBet series. Next I will write about the constraint that decided the rest of the executor: settlement is RSA-signed, so Cipher cannot invent a win.

If you want the same shape of engagement on another brand, book a coverage call on ciphersolutions.dev.

FAQ

Who is writing this?

Kristiyan Radev, founder of Cipher Solutions. Founder-led Release QA for iGaming, based in Bulgaria. VoltageBet is the client on this B2B QA contract.

Is this a staff-aug seat?

No. VoltageBet contracted Cipher to own QA: coverage, automation, and a written verdict. I work in their TestRail and their sprint. They do not interview a bench.

Why not 181 green TestRail results?

Because the sportsbook provider signs settlement. Forging a win tests Cipher’s stub, not VoltageBet’s stack. Fake skips would have made a prettier nightly and a worse sign-off.

Where should I start if I hire Cipher?

The homepage is the offer. This log is what that offer looks like on a live sportsbook suite.