Inspiration
OneDoor was inspired by a simple problem: in Canada, help often already exists, but finding and navigating that help can be incredibly difficult.
Someone who loses their job, falls behind on rent, arrives as a newcomer, develops a disability, or faces another major life change may need to navigate several disconnected systems at once:
- Federal benefits
- Provincial programs
- Municipal services
- Food support
- Housing organizations
- Legal clinics
- Community agencies
The problem is not always a lack of resources. The problem is that people are expected to understand the system before the system can help them.
As a solo hacker, I wanted to change that.
Instead of building another directory, I asked:
What if someone could tell their story once, and the system could build a clear path forward?
That became OneDoor Canada.
What it does
OneDoor Canada turns a person's situation into a personalized, prioritized Escape Plan.
A user can simply describe what is happening in plain language:
"I lost my job, I have two children, I'm behind on rent, and I only have $800 left."
OneDoor then organizes possible next steps into an easy-to-follow timeline:
Today → Within 48 Hours → This Week → Next
Instead of giving users a long list of programs, OneDoor answers a more important question:
"What should I do first?"
Key features
- Life Event Engine for job loss, housing instability, newcomer support, disability, separation, student hardship, caregiving, and more
- Community Passport so users can tell their story once instead of repeatedly entering the same information
- Personalized Escape Plan
- Dependency-based sequencing that determines which steps should happen first
- Deterministic eligibility pre-screening for programs such as Employment Insurance
- Secure document upload and scanning
- Recovery progress tracker
- Consent-based selective sharing
- Privacy export and account deletion
- Dark and light modes
- Responsive mobile and desktop design
- Voice-ready interaction
- Multilingual-ready architecture
- Accessibility support
- Links back to trusted government and community sources
The core idea is simple:
One story. One plan. No wrong door.
How I built it
I built OneDoor as a full-stack web application.
Backend
The backend uses:
- Python
- FastAPI
- Secure authentication
- HttpOnly sessions
- CSRF protection
- Password hashing
- Encryption for sensitive stored information
- Consent and audit logging
- Versioned eligibility rules
- Secure document validation
- File hashing
- Encrypted document storage
- Optional malware scanning
- Privacy export and deletion controls
Frontend
The frontend includes:
- Responsive navigation
- Canadian-inspired visual design
- Animated backgrounds
- Glow effects
- Dark and light mode
- Life-event selection
- Community Passport
- Escape Plan generation
- Dependency visualization
- Recovery progress tracking
- Privacy controls
- Accessibility features
- Reduced-motion support
Responsible AI architecture
One of the most important design decisions I made was separating AI-style language understanding from eligibility decisions.
AI can help with:
- Understanding natural language
- Translation
- Simplifying complicated government wording
- Document interpretation
- Identifying missing information
But deterministic rules handle:
- Eligibility calculations
- Income thresholds
- Age requirements
- Residency requirements
- Deadlines
- Program dependencies
I did not want an AI model making high-impact decisions that should be transparent, traceable, and based on authoritative rules.
Challenges I ran into
One of the biggest challenges was realizing that simply building a better benefits search engine would not be enough.
Canada already has tools that help people discover benefits and services.
The deeper broken system is coordination.
People do not experience problems in isolated government categories.
Someone who loses their job may simultaneously need:
- Income support
- Housing assistance
- Food assistance
- Childcare support
- Employment services
- Legal help
That pushed me from a search-based model toward a journey-based model.
Another major challenge was eligibility.
Using an AI model to make definitive decisions about government benefits would introduce serious risk. I therefore created a deterministic, versioned rules engine and clearly label results as pre-screening guidance rather than official government decisions.
Privacy was another major challenge because OneDoor may handle highly sensitive information.
I designed the Community Passport around data minimization, allowing users to choose exactly what information they want to share.
Document security was also more complex than allowing simple uploads, so I added:
- File-type validation
- File signature checks
- Content inspection
- Hashing
- Encrypted storage
- Suspicious PDF detection
- Archive checks
- Optional malware scanning
Accomplishments that I'm proud of
I am especially proud that OneDoor became much more than a chatbot or benefits directory.
I built a system that can take a complicated life situation and turn it into an actionable sequence of next steps.
The feature I am most proud of is:
Tell Your Story Once
The Community Passport lets someone enter important information once and reuse it throughout their journey while still controlling what gets shared.
I am also proud of the dependency-based Escape Plan, which helps users understand not only what resources exist, but also what order they should approach them in.
As a solo hacker, I designed and built:
- The full-stack application
- Frontend experience
- Backend architecture
- Secure authentication
- Encrypted storage
- Deterministic eligibility rules
- Document security workflows
- Consent history
- Selective-sharing controls
- Privacy features
- Responsive interface
- Dark/light mode
- Accessibility support
- Deployment configuration
Building all of these pieces myself was one of the most rewarding parts of the project.
What I learned
The biggest thing I learned is that a broken system does not always mean that nothing exists.
Sometimes the individual pieces work, but the connections between them do not.
Canada has many valuable:
- Government programs
- Nonprofit organizations
- Community services
- Information portals
But someone experiencing a crisis may still have to navigate all of them independently.
I also learned that responsible AI design requires knowing when not to use AI.
AI can be extremely useful for understanding and simplifying information, but high-impact decisions should remain:
Deterministic → Transparent → Traceable → Source-based
I also learned how important privacy-by-design is when building technology for vulnerable communities.
Instead of asking:
"How much data can I collect?"
I designed OneDoor around:
"What is the minimum information needed to help this person complete this particular step?"
That principle shaped much of the application.
What's next for OneDoor
My next step is turning OneDoor from a strong working prototype into a platform that could safely operate across Canada.
Future development
- Integrate more federal programs
- Expand provincial and municipal coverage
- Add more nonprofit and community services
- Expand deterministic eligibility engines beyond Employment Insurance
- Add live geospatial matching for nearby services
- Build secure referral workflows
- Add deadline and follow-up notifications
- Expand document intelligence
- Add form preparation assistance
- Complete professional French translation review
- Expand multilingual support
- Conduct formal accessibility testing with users and assistive technologies
- Complete external privacy and security reviews
- Move encryption keys to managed cloud key infrastructure
- Add production-grade cloud database infrastructure
- Build dashboards for participating service providers
- Explore partnerships with municipalities, nonprofits, settlement organizations, and community agencies
My long-term vision is larger than helping people find benefits.
I want OneDoor to become an orchestration layer between people and fragmented public services.
Because when someone's life is falling apart, they should not have to become an expert in government systems.
I didn't build another door into the system.
I built the way through it.
Built With
- accessibility
- ai
- api
- authentication
- clamav
- cryptography
- css3
- cybersecurity
- encryption
- fastapi
- government
- html5
- httpx
- javascript
- privacy
- processing
- pwa
- python
- responsive
- rest
- speech
- uvicorn
- wcag
- web
Log in or sign up for Devpost to join the conversation.