Business instinct.
Builder's mindset.

My contribution. I designed and built the website and the member commerce platform using agentic coding tools, from customer journeys through to delivery and release checks.

The difficult part. A successful checkout has to lead to the correct access, every time. The payment flow checks the signature, handles repeat events, and corroborates provider state before it opens an entitlement.

Evidence. A 9,014-line pre-deploy gate checks every release, with live probes and automatic rollback after each deploy. Payments went live in August 2026: shipped capability, not large-scale customer traction.

My contribution. I built and deployed this storefront by myself, connecting the entire purchase journey: checkout, payment verification, transaction records, and delivery.

The difficult part. Fixing roughly 60 defects in one launch day meant checking the whole path from a customer's click to their download link, not just the parts that looked finished.

Evidence. Earlier, I ran a one-day launch audit for a family venture, Flippy Tales. That habit, checking a launch end to end before calling it done, carried straight into my own storefront.

My contribution. I designed the five-phase pipeline and pinned its computation to specific versions, so the same inputs produce the same outputs every time it runs.

The difficult part. Repeatability is harder than raw output. Each phase carries an explicit contract, and nothing publishes until the verify phase passes.

Evidence. The documented estate includes 15,117 Pine Script files, and the pipeline runs unattended on daily and weekly schedules.

My contribution. I built the daily publication gate: a fail-closed check that stops output whenever something does not pass.

The difficult part. Explaining market structure without implying a trading signal. The illustration uses synthetic data and shows how one instrument looks across five horizons at once; it is not a live market reading.

Evidence. 212 instruments, 17 sectors, and a gate that has to pass before anything publishes.

My contribution. I designed the split between judgment and mechanics: agents that decide what to say, and agents that only execute and measure.

The difficult part. Keeping twelve platforms consistent without twelve separate improvisations. The distribution map and the attribution convention are what make the output comparable week to week.

Evidence. The dashboard image shown here uses placeholder figures. It demonstrates the operating workflow, not campaign results.

My contribution. I designed the studios' script gates, scene maps, and quality checks, so a written teaching method could become watchable video without losing editorial control.

The difficult part. Thirty-one agents support scripting, scene maps, narration, subtitles, and quality checks across two studios. Keeping that many handoffs bounded, so nothing renders without a human gate, took real structure.

Evidence. The 72-chapter flagship course is built and QA-passed, and the book runs across five channels plus an India edition.

My contribution. I turned one written teaching method into several formats: a book, a video course, and later an audiobook script, while keeping the same argument intact across all of them.

The difficult part. A book and a video course teach differently. Restructuring the same material into 72 chained chapters, each two to three minutes, without losing the thread, was the real work.

Evidence. The India edition adds a 152-page NIFTY and MCX localisation. Both editions went through the same editorial gate before publication.

My contribution. I built this in a single working session: one HTML application plus a small reproducibility script, with no external dependencies.

The difficult part. Time and tool spend are usually invisible costs. A separate token-cost auditor prices every session against published token rates, so I can see the real cost of the work, not just a usage bar.

Evidence. 2,581 founder hours reconciled, with every dated AI-tool cost line kept alongside them.