AI Developer (Contract) — Menu Intelligence & Vision AI

FabTechSol; Pakistan; Remote; PKR 26,757,000 - 57,336,000; Part time

Job summary

A tech solutions provider is seeking an AI Specialist to improve extraction accuracy for digital menu creation. You will work with a dedicated team focusing on AI-driven image processing to recreate menus into editable formats. The role requires proven hands-on experience with vision AI models and a keen eye for design detail. This is a contract role with immediate availability required, allowing you to work remotely from any timezone.

Qualifications

• Real, hands-on experience with vision AI models in production.
• Experience handling real-world PDFs, photos, and scanned documents.
• Ability to accurately identify color hex and RGB values.
• Experience detecting columns, hierarchy, spacing, and element positioning.
• Strong eye for design fidelity and output fidelity.

Responsibilities

• Ensure extraction accuracy and fidelity of the AI layer.
• Iterate to improve on edge cases during document processing.
• Collaborate with frontend/backend teams to deliver editor-ready data structures.

Skills

Vision AI models (GPT-4o Vision, Claude Vision, or equivalent)

Image and document processing

Color extraction and analysis

Layout detection

Eye for design accuracy

Tools

OpenCV

Job description

Overview

Use vision AI to recreate restaurant menus as pixel-perfect, fully editable digital versions. Contract role. Immediate start.

Type: Contract • Project-based

Location: Remote • Any timezone

Start: Immediate • Available now

The Project

We are building a platform where a restaurant owner uploads their menu — and our AI recreates it as a pixel-perfect, fully editable digital version.

Same colors. Same fonts. Same decorators. Same logo. Same layout. Same positioning of every single element. Just editable.

The result should feel like the original menu was lifted off the page and placed into an editor — not approximated, not templated. Reconstructed.

You will not be working alone. A dedicated frontend team and backend team are already in place. Your focus is purely the AI layer — making the extraction smarter, more accurate, and more faithful to the original design.

The Specific Challenge

When a menu image or PDF is uploaded, the AI needs to handle all five of these accurately

• Identify and extract every color used — background fills, text colors, border tints — down to accurate hex values. Not approximate. Exact.
• Logo Capture — Detect the restaurant's logo — its position on the page, its bounding region, its design style — and represent it faithfully in the output structure.
• Identify borders, separators, ornamental lines, and graphic elements — and capture their exact positions, sizes, and styles relative to the page.
• Structured Text Extraction — Read all text content — section headers, item names, descriptions, prices — and return it in a clean, structured format that preserves the original grouping and hierarchy.
• Layout Reproduction — Understand and reproduce the column structure, spacing, element hierarchy, and overall composition — so the digital recreation matches the original without guesswork.

Your Scope — AI Layer Only

Your Focus

• Extraction accuracy & fidelity
• Iterative improvement on edge cases

Required Skills

Real, hands-on experience only — we verify during shortlisting

• Vision AI models (GPT-4o Vision, Claude Vision, or equivalent) — production use
• Image and document processing — real-world PDFs, photos, scanned documents
• Color extraction and analysis — accurate hex/RGB identification
• Layout detection — columns, hierarchy, spacing, element positioning
• Eye for design accuracy — output fidelity matters to you, not just partial results

Bonus Skills

Stand out if you have these

• Computer vision / image segmentation (OpenCV or equivalent)
• Logo detection and classification
• CSS/HTML awareness — understanding how output will be rendered in an editor
• Restaurant, print design, or digital menu industry experience

What Good Looks Like

You upload a photo of a handwritten chalkboard menu and the AI returns the exact background color, all text structured by section, and the ornamental border accurately positioned.

You upload a scanned PDF of a fine-dining menu with a gold logo and the AI extracts the logo region, the gold hex values, and every item grouped under its correct course heading.

We respond to all shortlisted candidates within 24 hours. Serious applicants may be given a small paid test task before the contract starts.

Apply for This Role

Full Name *

Email *

Phone

WhatsApp

Years of Experience *

Expected Rate

LinkedIn Profile

Required Skills — Check all that apply *

Only check skills you have real, hands-on experience with.

• Vision AI models (GPT-4o Vision, Claude Vision, or equivalent) — production use
• Prompt engineering — structured, deterministic, reliable JSON output
• Image and document processing — real-world PDFs, photos, scanned documents
• Color extraction and analysis — accurate hex/RGB identification
• Layout detection — columns, hierarchy, spacing, element positioning
• Eye for design accuracy — output fidelity matters to you, not just partial results

Bonus Skills — Optional

• OCR pipelines (Tesseract, AWS Textract, Google Document AI)
• Computer vision / image segmentation (OpenCV or equivalent)
• Logo detection and classification
• CSS/HTML awareness — understanding how output will be rendered in an editor
• Restaurant, print design, or digital menu industry experience

How Would You Approach This Challenge?

This is the most important field. Be specific about models, prompt strategies, and how you'd handle edge cases like low-quality scans or complex layouts.

Describe a Similar Project You've Built

Additional Notes (Optional)

CV / Resume (Optional)

Click to upload CV — PDF, DOC, DOCX, max 10MB

I am available to start immediately on a contract basis *

This role requires immediate availability. Shortlisted candidates will be contacted within 24 hours.