Brief description of the organization
Founders Ayesha Khan and Kash Dimas are launching an AI/ML-powered online interior decorating business that generates photorealistic, personalized room designs for clients based on their style preferences, room dimensions, and budget. Users receive AI-designed rooms with curated, shoppable furniture selections they can purchase directly through their platform.
This Capstone engagement invites a University of Waterloo student team to assess, extend, and complete the founder’s platform, delivering a fully functional, launch-ready product by the end of the semester (we are open to extending to a second semester, if required). Note: some code has been built and we anticipate this project will require one semester, however, we are open to spanning two semesters if warranted.
Students joining this engagement work directly with founders who bring 20+ years of combined business, marketing, and education experience. The founding team is committed to weekly check-ins, active mentorship, and supporting students in developing not just technical skills but the strategic and scalable thinking that real-world product development demands including navigating workplace nuances.
Problem area
The central problem students will explore is the technical integration and commercialization of a generative AI interior design platform.
While a baseline AI model exists in isolation, the core challenge lies in transforming it into a launch-ready product by solving three primary complexities:
Intelligence & Accuracy: Transitioning the AI from generic rendering to precise, rule-based design. Students must train the model to follow strict interior design logic (spatial relationships, material cohesion, and lighting) while matching designs to real-world, shippable inventory.
Systems Architecture: Building the connective tissue between disparate systems—integrating the AI engine with a scalable SQL-based inventory database, a multi-user dashboard ecosystem, and a functional e-commerce checkout flow.
Commercial Integrity: Solving real-world business constraints, such as implementing geographic shipping logic, protecting intellectual property through image/code obfuscation, and minimizing AI hallucinations to ensure the furniture rendered is exactly what user/customer receives.
Essentially, the team is tasked with taking a proof-of-concept and engineering the robust, secure, and scalable backend required to support a live business.
Main objectives
The main objectives of this project centre on transitioning the platform from a fragmented codebase into a fully integrated, secure, and market-ready commercial product.
The specific goals are:
- AI Sophistication and Alignment: Train the generative model to move beyond pretty pictures to designs that follow strict professional design rules (material matching, spatial flow, and lighting) while ensuring 1:1 accuracy between AI renders and actual supplier inventory and the client’s brand.
- Infrastructure Transformation: Replace the current static CSV-based inventory with a scalable database architecture and build a cohesive system that links the AI engine to the frontend, e-commerce checkout, and supplier logistics.
- Business Logic Integration: Implement real-world constraints into the code, such as geographic shipping eligibility (GTA-only restrictions) and budget-aware product selection.
- IP and Revenue Protection: Develop a suite of security features to safeguard the founders' intellectual property, including code obfuscation and advanced image protection (watermarking, screenshot deterrence, and disabling "save-as" functions).
- Operational Dashboarding: Build a three-tier dashboard ecosystem—User, Supplier, and Admin—to allow the founders to manage inventory, track sales analytics, and coordinate with logistics partners.
Ultimately, the objective is the successful handoff of a stable production environment that is ready to accept real customers and process transactions immediately upon project completion.
Scope of work
The project is structured into three strategic phases that move the platform from technical assessment to a secure, commercial launch.
Phase 1: Technical Audit & AI Core Integration
The foundation of the project involves stabilizing the existing codebase and teaching the AI to think like a designer.
- Codebase Assessment: Conduct a deep-dive audit of the inherited backend and AI models to identify gaps and technical debt.
- Design Logic Training: Train the AI using the proprietary Design Manual & Core Integration guides to ensure it understands spatial flow, material matching (metals, woods, fabrics), and lighting.
- Database Migration: Transition the inventory system from a static CSV to a scalable, automated database architecture.
- Constraint Programming: Integrate real-world logic, such as mapping supplier shipping zones (e.g., GTA-only) and user budget parameters.
Phase 2: Systems Architecture & E-Commerce Build
This phase focuses on building the functional business layer that allows for transactions and inventory management.
- E-Commerce Integration: Build the full purchase flow, ensuring the AI can lock in hero pieces and process orders through a secure payment gateway.
- CMS & Admin Development: Create a secure Content Management System so the founders can add or edit products without needing to touch the code.
- Feedback Loop Implementation: Build a machine learning loop where the AI learns from user refinements (swapping furniture) to improve future design accuracy.
- Cloud Deployment: Deploy the integrated system to the designated production environment for end-to-end testing.
Phase 3: Frontend Implementation & IP Fortification
The final phase polishes the user experience and secures the platform’s competitive advantages.
- UI/UX Development: Implement the provided wireframes to create the interactive questionnaire and the final purchasable room display.
- Dashboard Ecosystem: Finalize three distinct interfaces for Users (past designs), Suppliers (logistics), and Admins (sales analytics).
- IP Security Engineering: Apply "Hardened" protections, including code obfuscation, watermarking, and technical deterrence against screenshots or image saving.
- End-to-End QA: Conduct rigorous testing to ensure the rendered designs perfectly match the shippable products and dimensions.
Deliverables
The deliverables for this project bridge the gap between technical assessment and a market-ready deployment
For Problem Definition (Assessment Phase)
- Technical Audit & Codebase Baseline: A formal document or presentation detailing the status of the inherited code, identified gaps, and the approach to continues the build of the AI model.
- AI Training Logic Specification: A technical mapping showing how the Design Manual & the Core Integration rules (spatial flow, finish matching) are translated into the AI’s decision-making parameters.
- Database Schema & Migration Map: A blueprint for the transition from the CSV file to a scalable, automated inventory database.
Solution Verification A (Technical Validation)
- Automated Testing Suite: A set of tests that verify the Supplier Logic works (e.g., ensuring a user outside the GTA isn't shown GTA-only furniture) and that the e-commerce flow is secure.
- AI Model Validation Report: A comparison of AI-generated room renders against your sample room renders to prove the photorealism and spatial accuracy meet your standards.
- Security & IP Obfuscation Audit: Verification that the code is minified/obfuscated and that image protections (screenshot deterrence, no-save scripts) are active and functional.
Solution Validation B (Commercial Readiness)
- Fully Integrated Production Environment: The live platform deployed to our Google Cloud infrastructure, containing all frontend wireframes and backend logic.
- Triple Dashboard Ecosystem: Functional portals for the Admin (analytics), Supplier (logistics), and User (design history).
- ML Feedback Loop Documentation: A summary of how the system tracks user refinements to improve furniture selection intelligence over time, apply AI/ML learning to questions, supplier product catalogues, and scalable to add products and added shipping locations (local/global).
- Final Source Code Repository: Complete handoff of all original, tested code to our designated repositories.
Team meeting frequency
Weekly.
Skills and training required
With our project, the students aren't just coding, they are learning how to operate as technical consultants and product owners with a real-world project.
- Holistic Problem Solving: Managing the big picture by auditing an unknown codebase and architecting a cohesive system where AI, e-commerce, and logistics work in sync.
- Critical Thinking & Logic Design: Translating abstract interior design principles (e.g., modern luxe, artful eclectic) into strict, code-based rules for spatial accuracy and material pairing.
- Professional Presentation & Communication: Engaging in weekly founder check-ins and milestone reviews to explain technical progress, gaps, and solutions to non-technical stakeholders. Presenting findings similar to real-world presentations and obtaining feedback to improve with real-world experiences from seasoned leaders.
- AI/ML & Full-Stack Integration: Gaining hands-on experience deploying generative models and building scalable backend architectures in a live production environment.
- Cybersecurity & IP Strategy: Implementing sophisticated code obfuscation and digital asset protections to safeguard commercial intellectual property.
- Strategic Adaptability: Navigating the ambiguity of a startup environment, where students must solve real-world constraints like geographic shipping limits and budget-aware logic.
Resources required
The following technical and strategic resources will be provided to ensure the team can deliver a production-grade solution. Beyond these assets, the founding team is deeply committed to the students’ professional growth, offering market-ready mentorship and soft-skill development with real-world examples tailored to the team's specific career goals.
- Cloud Infrastructure & Version Control: Full access to the existing codebase hosted on GitHub and the production/staging environments on Google Cloud Platform (GCP).
- Proprietary AI Training Assets: Two comprehensive training manuals:
- Manual #1 (Design Logic): Proprietary rules for spatial accuracy, material pairing, and style-matching.
- Manual #2 (Technical Blueprint): Logic for automated inventory extraction and catalog management.
- Product & Supplier Data: A complete SKU-mapped inventory (CSV and supplier catalogs) including dimensions, pricing, and geographic shipping constraints.
- Executive Mentorship & Subject Matter Expertise: Weekly access to the founders for strategic guidance. Students will draw on 20+ years of leadership in marketing, digital products, interior/architectural design, and professional education to ensure the technical build aligns with real-world business goals.
- Design Assets: Full UX/UI wireframes and brand guidelines (logos, typography, watermarking specs), allowing the team to focus entirely on implementation and engineering.
- Development Tools: Access to and support for standard AI/ML frameworks (e.g., PyTorch, TensorFlow, or Stable Diffusion libraries) and full-stack tools compatible with the existing backend.
NDA or a commercialization agreement for this project?
Yes