Zubair TanveerE-commerce Manager

The problem

PITB, the Government of Punjab's IT agency (est. 1999), delivers the Punjab E-Commerce Initiative. Its four E-Commerce Readiness Hubs help SMEs, artisans and home-based sellers get ready to sell online, free of charge. Each business gets product photography, creatives, social media content and an audit, branding, and online-store guidance. Many start with little online presence: in the hub's audits, 70% had low or no social media engagement and only 16% had a website (as of 2 Oct 2026).

Every business goes through intake, a shoot, product images, social posts, reels, an audit and a completion report, and five people share that work. The team tracked progress in a spreadsheet while the files lived in Google Drive, and the two often disagreed. My own manual audit on 2 Oct 2026 logged 75 discrepancies.

What I did

• Ran the hub: I lead the team through intake, production, sign-offs and delivery. I standardized intake (online registration plus a 15-point social media audit), I report to PITB management, and I have delivered 8+ e-commerce workshops.
• Role-based work queues: each person opens "My Work" and sees their next actions, such as starting a shoot, signing off photos, logging a follow-up or marking "products arrived".
• 21 data-quality rules: they compare the team's tracking sheet with Google Drive and send each mismatch to the person who owns it.
• Automatic completion reports: a branded 2-page PDF when a business completes, with a new version 30 minutes after any later change. Shareable weekly and monthly reports for management, plus Excel export.
• Privacy by design: national ID (CNIC) data is never stored, contact details are limited by role, and shared reports carry no contact data.
• Automations I control: I replaced a failing Make.com form-to-Drive automation with Drive folder setup built into the portal, and added a switchboard of 9 automations (7 can be turned on or off).

How I built it

1. Product and requirements: the roles, the rules a business must pass to count as complete, and what must never happen. The plan was approved on 2 Oct 2026.
2. System design: Google Drive and the team's sheet stay the sources of truth. The portal syncs Drive every 2 minutes and the sheet every 10 minutes.
3. Implementation: built phase by phase from the approved plan.
4. Testing: 950+ passing automated tests, end-to-end tests at phone and desktop widths (Playwright) and WCAG 2.1 AA accessibility checks. My manual audit became the acceptance test for the data-quality rules.
5. Launch and operations: live since 3 Oct 2026, and I run it in production.

Results

• 100+ businesses onboarded (as of Oct 2026), and 45 completed the full readiness package (as of 16 Oct 2026). Median of about 5 weeks from first visit to completion (as of 9 Oct 2026).
• 4,400+ files delivered: 2,960 product photos, 1,003 product images, 316 social posts, 71 reels and 63 branding designs (as of 16 Oct 2026).
• 88 online registrations in 5 weeks (25 Aug – 30 Sep 2026).
• Social media help (audit and guidance) delivered to 85 of 93 businesses (as of 9 Oct 2026).
• 950+ passing automated tests on the portal (as of 16 Oct 2026).
• The data-quality rules caught every reproducible finding of the 75-item manual audit (0 missed).
• 45 completion reports live, with 51 versions generated automatically (as of 9 Oct 2026). 167 deliverable sign-offs were logged in the first 5 working days after launch.

These are aggregate program figures only, with no data about individual businesses. PITB is named only to describe my role; this case study is not endorsed by PITB.

The brief

Xelara Beauty sells across five categories: makeup, skin care, hair, body and men's care. The brand needed a storefront that feels premium and editorial, keeps a launch catalog of 100+ products easy to browse, and sells in more than one country. The brand's Figma designer created the visual design; my job was to turn those designs into a working Shopify store that matches them closely.

Key decisions, in plain English

1. Start from Dawn, customize where it shows. Dawn is Shopify's own reference theme, so the cart, product pages and collection filters start from code Shopify maintains. The custom work went into what shoppers notice first: the homepage, the header and the product rows.
2. One product slider, set up 4 ways. The Bestsellers, Foundation and Concealer, Moisturizer & Serum and Lip Essentials rows are one reusable section with different settings, not 4 one-off sections. The next row is a few clicks in the theme editor, not new code.
3. Custom styles kept apart from Dawn's. The custom sections keep their styles in their own files and use an "xel-" class prefix, so they don't collide with Dawn's own styles, and future Dawn updates are safer to apply.
4. A mega menu for a big catalog. With 36 collections, a plain dropdown would bury most of the range. "Explore Collections" opens one panel with Eyes, Face Care, Skin Care and Lips, plus Fragrance, Body Care, Hair Care and Men's Care.
5. Local prices in each market. Shopify Markets shows prices in the shopper's currency, such as US dollars in the US and UAE dirhams in the UAE, and a country selector in the header lets shoppers switch.
6. No app bloat. The sliders, the menu and the other homepage features are part of the theme, so the homepage loads no third-party app scripts: only the store's own files and Shopify's, including Shop Pay.

What shipped

• A live store with a launch catalog of 103 products, 301 variants and 425 product images in 36 collections (as of 17 Oct 2026).
• 9 custom section types on the homepage, including the hero, the category grid, the reusable product slider (used 4 times), about and why-us blocks, and an Instagram feed.
• A custom header with the "Explore Collections" mega menu and a country selector.
• Product and collection pages in the brand's style: photo shade swatches (14 on the Lip Gloss page), filters, sorting and breadcrumbs, with Add to cart and Shop Pay.
• Local-currency prices for each market, through Shopify Markets.
• No third-party app scripts on the homepage (checked 17 Oct 2026).

Xelara Beauty is a client project, shown with the client's approval. Visual design by the brand's Figma designer; store build by me. The store facts here were checked on the public storefront on 17 Oct 2026. Sales and traffic figures belong to the client and are not shared.

The problem

Some Shopify merchants need an alternative payment provider, including merchants in countries where Shopify Payments isn't available. They still want a branded checkout, money paid into their own account, and real paid orders in Shopify, so their order and fulfillment workflow keeps running. Doing this safely is mostly a money problem: the payment and the order live in two systems, and payment notifications get re-sent.

What I did

I built and launched it as a self-serve platform: merchants sign up, connect their store and payment account, and pay per store. There's a public demo, and the public launch is planned for 31 Oct 2026.

Product and system-design decisions, in plain English

1. The money goes straight to the merchant. The shopper pays into the merchant's own payment account, and the paid order is created in Shopify automatically, so fulfillment and the merchant's other apps keep working.
2. Duplicate-proof settlement. Payment platforms re-send notifications, and requests get retried. Each payment can be claimed only once, so a re-sent notification or a retry can never create a second order.
3. Nothing stuck goes unnoticed. A 15-minute reconciliation alerts on any payment without an order, so a person can resolve it.
4. Ad tracking per merchant (my PPC background). Each merchant gets their own Google Ads conversion tracking for add-to-cart, begin-checkout and purchase, with the ad click carried over from the store to the checkout.
5. Onboarding without an operator. A self-serve 5-step wizard. Payment setup went from 5 copied values to 1 API key.
6. Growth built in. An affiliate program with 30-day first-click attribution.

How I built it

I wrote the requirements and the rules the money path must never break, starting with "no duplicate orders", and I made the system-design calls above. The build includes an automated test suite that checks those rules. I run the platform: releases, reconciliation alerts and fixes.

Results

• Live, with self-serve sign-up, a public demo and per-store subscription billing. Public launch planned for 31 Oct 2026.
• Duplicate-proof payment-to-order settlement: re-sent notifications and retries can't create a second order.
• 15-minute reconciliation that alerts on any payment without an order.
• Per-merchant Google Ads conversion tracking for add-to-cart, begin-checkout and purchase.
• Self-serve 5-step onboarding; payment setup cut from 5 copied values to 1 API key.
• An affiliate program with 30-day first-click attribution.

The problem

Merchants and dropshippers who import products need listings that are accurate, SEO-ready and in their own brand voice. AI writes them fast, but it can also invent facts, overwrite good data and get expensive per listing. To work as a product, every listing has to be correct, easy to undo and cheap to generate.

What I did

• Built and launched Listrova: from a product link to a reviewed draft in Shopify, with a one-click revert.
• Safe by default: products go to Shopify as drafts first, and every change can be reverted exactly. Credits are charged only when a product actually reaches Shopify, and a retried request is never charged twice. Store access tokens are encrypted.
• Designed a test-set evaluation: a fixed set of 16 test products (the "golden set"), each scored against 7 hard checks, to compare AI models and settings on accuracy, cost and speed.

How I built it

I defined the listing pipeline, the hard checks every listing must pass and the cost target, then built the app against them. The golden set was the testing gate: it decided which model and settings shipped.

Results

• 16/16 golden-set products passed all 7 hard checks (100% hard-check pass rate).
• About $0.03 per listing, with a median of about 13 seconds per listing.
• About 70% lower AI cost than with my first model choice (about $0.11 per listing).

The problem

Merchants want better product titles, descriptions and meta tags for search. But rewriting them in bulk with AI is risky: the AI can invent product features, and the changes are hard to undo. A Shopify app also has to meet the App Store's requirements before it can be listed, from billing to privacy.

What I did

• Built ListingLift, an app inside Shopify admin that rewrites titles, descriptions and meta tags from each product's own data, with an exact one-click Undo.
• Billing through Shopify: merchants buy credit packs through the Shopify Billing API.
• Fair credit billing: a store can never spend more credits than it has bought, and failed AI jobs are never charged.
• App Store readiness: Shopify's privacy (GDPR) data requests and app uninstalls are handled and verified, and I checked the app against 31 Shopify App Store requirements.

(opens full size in a new tab)

How I built it

I chose the problem and wrote the requirements and the App Store checklist, then built the app on Shopify's official app template with the Shopify CLI. QA ran against the deployed app, and the requirements check was the final gate.

Results

• Checked against 31 Shopify App Store requirements: 30 met, 0 failed, 1 still needs a manual check.
• Failed AI jobs are never charged, and every change can be undone exactly.
• Credits can't be overspent: in a test, 20 simultaneous credit requests against a balance of 10 got exactly 10.