The Architect of Impossible Solutions | Enhanced by AI
TCG OCR Scan & Shopify Integration
Problem:
Listing large Pokémon card inventories manually was slow and error-prone. Each card required manual input for name, set, etc.
Solution:
We developed an OCR-powered system that identifies TCG cards from scans (Document Scanner, Pictures with Smarpthone), matches them to a database. The system uses Tesseract OCR & Azure Computer Vision for image recognition.
Key Features:
• Dual OCR pipeline (Tesseract + Azure Vision)
• Automated card lookup via TCG API
• Review queue for low-confidence scans
• Auto-listing to Shopify with title, rarity, set & price
• Error handling and manual approval interface
Results:
• 97% faster product listing workflow
• Human review ensures 100% data accuracy
• Supports bulk imports for large collections
• Greatly improved speed and consistency in catalog updates
16
566
To pay my suppliers in different currency's, countries was always a pain. 150 invoices per month, manually... Wise is great to save on cost, but each invoice has to be manually entered or scanned.
BatchPayer solves the issue. It scans up to 1k invoices in bulk, identifies by OCR the data for the transfer and puts them into the batch file format.
Then you can export the CSV or push it by API to WISE to wire it.
Working:
👍 OCR Scan - still improving multi language, currency etc.
👍Review System (Check if all infos are extracted correctly)
👍Payment System
Working on:
🧠 API integration (working on compliance rules etc. as sending money by wire transfer is critical)
🧠 Smartphone App to add Scanner
🧠 Own extraction model & refine existing one by mismatch etc.
Project Link (https://invoice-ocr-upload-for-wise-362.created.app/)
5
435
Laser Comparison Platform
Problem:
Finding the right laser engraver or cutter was difficult for me. Specs, prices, and models were spread across many manufacturer sites. I couldn’t easily compare key details like laser type, power, wavelength, or supported materials.
Solution:
I build a powerful comparison and search platform that helps users identify lasers matching their needs. The site combines a Bolt.com frontend with Supabase for structured product data, while n8n and Firecrawl handle scheduled price checks and data validation to keep listings accurate and up to date.
Key Features:
• Over 100+ filter options
• Live database powered by Supabase
• Data validation & more via n8n + Firecrawl
• Custom compatibility scoring algorithm
Tools Used:
Bolt.com
Supabase
n8n
Firecrawl
Custom Code
17
547
Centralized Sales Automation System
🔍 Problem
The client launches dozens of new products each month. Every product had its own Zapier workflow. Managing hundreds of automations became slow, error-prone, and hard to maintain.
💡 Solution
We built a centralized automation system using n8n with a custom webhook plugin. All orders are now sent to a single webhook that references a Google Sheet for configuration.
Key Automations:
• Single webhook endpoint for all products
• Dynamic routing via Google Sheets
• Auto-updates to CRM, Kajabi, Discord...
• No-code management - changes made directly in Sheets
⚙️ Results:
🕒 Replaced 300+ Zaps with one scalable workflow
⚙️ 95% less maintenance
🚀 New launches need only one new row in Sheets
🧰 Tools Used:
n8n
Google Sheets
ActiveCampaign
Discord Apps
Custom Webhook Plugin
Forum API
Kajabi
AI SaaS Dev | LLMs, Agents, Voice & Automation | Web, Mobile
1x
Hired
5.0
Rating
6
Followers
AI SaaS Dev | LLMs, Agents, Voice & Automation | Web, Mobile
OCR Receipt Parsing Microservice (AI-Powered Backend System)
Most receipt-based systems fail because the data is messy, inconsistent, and spread across formats that machines don’t naturally understand. People don’t realise it, but the real problem isn’t capturing receipts, it’s turning them into reliable, structured data that can actually be used.
This system removes that friction entirely. You send a receipt (image or PDF), and it comes back as clean, structured JSON ready to plug into any workflow.
The core problem it solves:
Receipt data is chaotic. Different formats, inconsistent naming, missing structure, and OCR noise make it hard to extract anything usable. Even when OCR works, the output is raw text, not something you can build logic on top of.
This project builds a full processing layer that doesn’t just read receipts, it understands and standardises them.
What was built:
A backend microservice that acts as a structured data engine for receipts. The system accepts images or PDFs via an API, runs OCR, extracts merchant details, dates, totals, and line items, and converts everything into a strict JSON schema.
But the real value sits in what happens after OCR.
A normalization layer cleans and standardises item names so inconsistent inputs like “BANANA”, “Bananas”, or “Banana 1lb” all map to a single canonical item. Quantities and prices are cleaned, structured, and validated so the output becomes consistent across different stores and formats.
The system can also plug directly into Airtable, pushing structured items into a live database, enabling automated workflows like pantry tracking, expense logging, or analytics pipelines without needing a full backend system.
Everything is exposed through a simple /parse-receipt API, making it easy to integrate into mobile apps, SaaS products, or internal tools.
Technical architecture:
FastAPI-based microservice designed for simplicity and performance, with OCR powered by Tesseract or cloud services like AWS Textract and Google Vision depending on accuracy requirements. The parsing layer combines rule-based extraction with AI-assisted cleanup to handle real-world receipt noise.
The system is fully containerized using Docker, deployable on platforms like Render or Heroku, and comes with OpenAPI (Swagger) documentation for quick testing and integration.
Designed as a stateless service, it avoids database complexity and instead integrates with external systems (like Airtable), making it lightweight and easy to scale.
Business value built in:
This isn’t just an OCR tool, it’s a data standardization engine. The same system can power expense tracking apps, inventory systems, meal planning products, or financial analytics platforms.
Because the parsing and normalization layers are modular, the microservice can be exposed as a standalone API, creating opportunities for reuse across multiple products or even external licensing.
I built a gamified directory called hivebid.lol (https://hivebid.lol) where users bid money 🎉 trying to dominate the hive to promote their SaaS, Service, or simply their social media account.
AND It's driving me crazy, am so excited 😍
DocMorph AI is a full-stack web application designed to eliminate manual data entry. It ingests locked, scanned PDFs, runs advanced Optical Character Recognition (OCR), and reconstructs the data into clean, editable Microsoft Word documents.
The Challenge:
Standard OCR scripts frequently fail or crash the server when attempting to process large, multi-page PDFs due to massive RAM consumption. The client needed a lightweight, scalable solution that wouldn't time out during heavy workloads.
The Solution:
I architected a decoupled monorepo system. The frontend was built with React for a seamless, drag-and-drop user experience. For the backend, I engineered a custom Node.js/Express.js server utilizing Tesseract.js. To ensure stability, I developed a custom text-chunking algorithm that slices the PDF into manageable data streams, processes them safely, and stitches the final layout together into a downloadable .docx file.
0
105
UGC ads usually mean finding a creator, sending a brief, waiting a week, then paying per video. I wanted to see how much of that an n8n workflow could handle.
The video in this post was not filmed. It was generated end to end by a workflow I built.
How it works:
I drop in a product, a short brief and a target audience
An LLM writes the UGC-style script with a hook, the body and a call to action
Veo generates the video in a natural, creator-style look
The workflow publishes it to TikTok, Reels and Shorts [and schedules it for the best time]
So the whole path from brief to published video runs without me touching an editor.
What I learned:
The prompt matters more than the model. Specific lighting, camera and tone details made the difference between “obvious AI” and “believable creator”.
Hooks decide everything. I generate several hooks per product and test them.
Always label AI-generated content where the platform requires it. It’s the right thing to do and it avoids trouble later.
If you run a brand, an e-commerce store or an agency and want a steady stream of short-form content without the cost of manual production, I build these workflows. Message me and I’ll show you how it would work for your product.
Would you trust an AI-generated UGC ad for your brand? Curious where people stand on this.
#n8n #Veo #AIVideo #UGC #AIAutomation #ContentCreation #ShortFormVideo #Ecommerce #MarketingAutomation #GenerativeAI #NoCode #SocialMediaMarketing
1
1
205
🚀 Just shipped a project: BrandBuzz, a fully interactive digital marketing agency landing page built with React + Vite!
As a Computer Science student, I wanted to move beyond tutorials and recreate a real UI design in code, so I took a design mockup and turned it into a working, responsive site.
What I built:
⚡ Sticky navbar with smooth scrolling and scroll-based active link highlighting
🎠 Auto-playing testimonials carousel that pauses on hover
🧩 Modal forms for Sign Up, Get Started and Contact Us, with keyboard support (Esc to close)
✨ Scroll-reveal animations using the Intersection Observer API
📱 A mobile-first responsive layout with a hamburger menu
♿ Accessibility basics: focus states, aria labels and reduced-motion support
What I learned:
🔹 Managing UI state with React hooks (useState, useEffect) and writing a custom useReveal hook
🔹 Building layouts with CSS Grid and Flexbox
🔹 Cleaning up event listeners and intervals properly to avoid memory leaks
🔹 Optimizing images (WebP) for faster load times
The biggest takeaway is that the small details, like hover states, transitions and spacing, are what make a website feel polished and professional.
I’m actively looking for internship and entry-level opportunities in frontend and web development, so I’d love to connect with developers, designers and recruiters. Any feedback on the project is welcome. 🙌
0
22
⚖️ Just finished building a law firm landing page from scratch, and I’m pretty proud of how it turned out!
I took a UI design and created it in React + Vite as a project for a law firm. My goal was to match the design as closely as I could, down to the spacing, typography and colors.
Here’s what I worked on:
✅ A responsive layout that works on desktop, tablet and mobile
✅ Practice area cards, a testimonials section and a contact form
✅ A clean navy and gold palette for a professional, trustworthy feel
✅ Reusable components, so the code stays organized
The hardest part was getting the small details right, like the card dividers, the font sizes and the hero image positioning.
I spent more time on these than I expected, but it taught me how much small design choices matter. I’m still learning and improving every day, so I’d love your feedback! What would you add or change?
🙌 If you’re hiring for frontend or web development roles, or just want to connect, feel free to message me!
#ReactJS (https://www.linkedin.com/search/results/all/?keywords=%23reactjs&origin=HASH_TAG_FROM_FEED) #Vite (https://www.linkedin.com/search/results/all/?keywords=%23vite&origin=HASH_TAG_FROM_FEED) #FrontendDevelopment (https://www.linkedin.com/search/results/all/?keywords=%23frontenddevelopment&origin=HASH_TAG_FROM_FEED) #WebDevelopment (https://www.linkedin.com/search/results/all/?keywords=%23webdevelopment&origin=HASH_TAG_FROM_FEED) #JavaScript (https://www.linkedin.com/search/results/all/?keywords=%23javascript&origin=HASH_TAG_FROM_FEED) #UIDesign (https://www.linkedin.com/search/results/all/?keywords=%23uidesign&origin=HASH_TAG_FROM_FEED) #ResponsiveDesign (https://www.linkedin.com/search/results/all/?keywords=%23responsivedesign&origin=HASH_TAG_FROM_FEED) #LandingPage (https://www.linkedin.com/search/results/all/?keywords=%23landingpage&origin=HASH_TAG_FROM_FEED) #LawFirmWebsite (https://www.linkedin.com/search/results/all/?keywords=%23lawfirmwebsite&origin=HASH_TAG_FROM_FEED) #StudentDeveloper (https://www.linkedin.com/search/results/all/?keywords=%23studentdeveloper&origin=HASH_TAG_FROM_FEED) #LearningInPublic (https://www.linkedin.com/search/results/all/?keywords=%23learninginpublic&origin=HASH_TAG_FROM_FEED) #BuildInPublic (https://www.linkedin.com/search/results/all/?keywords=%23buildinpublic&origin=HASH_TAG_FROM_FEED) #WebDesign (https://www.linkedin.com/search/results/all/?keywords=%23webdesign&origin=HASH_TAG_FROM_FEED) #OpenToWork (https://www.linkedin.com/search/results/all/?keywords=%23opentowork&origin=HASH_TAG_FROM_FEED) #100DaysOfCode (https://www.linkedin.com/search/results/all/?keywords=%23100daysofcode&origin=HASH_TAG_FROM_FEED)
Developed a system that replaces traditional OCR by using AI to interpret full-page handwritten documents.
Instead of extracting text line-by-line, the system:
Identifies document type
Uses a structured JSON contract to define required fields
Sends full-page images to an AI model
Returns clean, structured data (including multi-record tables)
Supports:
Variable number of records per document
Field validation and normalization
Spreadsheet-ready outputs
This approach significantly improves accuracy compared to OCR-based solutions.
0
93
ARJAN — My Personal AI Consulting Platform, Built on Zo Computer
I'm Ali Beheshti — AI consultant, founder of iTAB, and professor of Data Science Ethics at George Washington University. ARJAN is where those three identities meet.
🔗 Live site: https://arjan-site-ab.zocomputer.io/
What I built on Zo:
→ Mission Readiness Score — visitors adjust 4 live sliders across Quality Engineering, Section 508, AI Governance, and Mission Security. An animated ring scores them in real time, identifies their top gap and strongest area, and one click sends the full assessment into my contact flow
→ Live TTS narration on every capability card with synced ASL overlay — Section 508 compliant, built for everyone
→ Gmail-powered contact workflow via Zo's native integration — 4 simultaneous email paths on every submission. No third-party tools. Just Zo.
→ Capability cards reshuffle on every visit — no two visits are identical
→ Pixel-aware logo hover animation with inline sound toggle
My philosophy: AI should free people to focus on what matters — not replace them. Every feature on this site exists to serve the visitor. That is the only reason to build anything.
#zocomputerchallenge
1
2
139
Designed a multimodal AI system combining voice input, document understanding, and structured data extraction.
The system processes full-page documents and spoken input together, allowing flexible and efficient data capture.
Key capabilities:
AI-based document interpretation (beyond OCR)
Voice-driven interaction and data entry
Structured JSON outputs for downstream systems
Support for multi-record extraction and validation
This approach enables fully automated workflows in data-heavy environments and field operations.
0
88
Developed an AI-powered platform for automating field data collection, analysis, and reporting.
The system integrates multiple input methods, including handwritten sheets, voice input, and direct data entry.
Key capabilities:
Automated data extraction and validation
AI-assisted report generation
Workflow tracking and project management
Customizable outputs for engineering and technical teams
Designed to reduce manual work and improve accuracy in field operations.
The goal was to balance strong visual storytelling with clear structure, smooth user flow, and conversion-focused layouts.
Full Case Study
https://lnkd.in/gMJHd6tE
1
2
115
Luxury Furniture E-Commerce Homepage design
Thoughts? 💡 Would appreciate hearing your take.