I taught commerce for 11 years. I've never written a line of code.
I'm Deepak Jha. I'm not a security engineer and I don't pretend to be one — I'm an accounts guy from Maharashtra who ended up building a security scanner because I ran straight into the exact problem it exists to catch.
How I actually got here
Three years in a CA articleship in Pune, then eleven years running my own commerce coaching classes — accounts, tax, finance, batch after batch. Teaching forces you to do one thing well: take something complicated and make it click for someone seeing it for the first time. Every finding StackSecured writes — what's wrong, why it matters, what to paste to fix it — is that same instinct, pointed at security instead of a balance sheet.
In 2026 I registered Bullwiser Analytics Pvt Ltd and built BullWiser, a mutual fund analysis platform, without writing a line of its code myself. I directed AI to build it — the way an auditor directs a junior: here's what I need, here's how the number must be calculated, here's what's wrong with your draft, redo it. The AI wrote the code. I decided what got built and what "correct" actually meant.
Directing AI instead of writing code myself taught me something the hard way: AI-generated code ships fast and looks finished, but it quietly skips security defaults a human engineer would catch out of habit. That's not a theory — I proved it on my own second product.
In August 2026 I ran BullWiser through StackSecured myself and it found a real one: an unauthenticated endpoint returning full reviewer email addresses to anyone who asked, no login required — live, on my own platform, that I'd have sworn was fine. I fixed it the same day and rescanned to confirm it actually closed, not just that I assumed it did.
That's exactly the gap StackSecured is built to close. Not a hypothetical in a sales page — I found it on my own site, with my own tool, before anyone else did.
Who builds what
What I bring
An outsider's instinct for exactly the corner a non-engineer would skip. A teacher's refusal to ship an explanation nobody can act on. And firsthand proof — I'm the target customer, not just the vendor.
What AI brings
The 33 scanning engines themselves — written and rewritten under direction, the way I'd direct any junior: here's what needs checking, here's what "honest" looks like, redo it until it is.
Non-negotiables
Four lines I won't cross.
No inflated findings.
If an engine could not test something this scan, the report says exactly that — "could not be tested" — instead of a guess dressed up as a clean pass.
No manufactured urgency.
A finding is rated by real impact, not by what sells a report. If it is Low, it says Low.
The score reflects what is actually true right now.
Fix something and rescan — the report tells you it is fixed, not just what is currently wrong. Findings that no longer reproduce are tracked as resolved, not silently dropped.
A trust badge has to be earned.
The embeddable "Secured" badge only renders for a domain with verified ownership and a real score above the bar. No verified scan, no score above the line — no badge. There is no version of this where the number is decorative.
Deepak Jha
Founder, StackSecured — a product of Bullwiser Analytics Pvt Ltd