Inspiration
Affordable housing compliance is one of the most underserved areas in proptech. Working on NYC's largest 100% affordable housing development in 40 years — Willets Point Commons (880 units) — I experienced firsthand how much time compliance teams spend manually reading income documents, calculating AMI percentages, and preparing certification files. Every applicant requires the same repetitive process: collect pay stubs, tax returns, and offer letters, then manually extract numbers, calculate household income, and determine eligibility against HUD income limits.
With Mayor Mamdani's Block by Block plan targeting 400,000 new affordable homes in NYC, this manual process simply cannot scale. CompliQ was built to solve this.
What it does
CompliQ is an AI-powered document intelligence platform for affordable housing compliance. It allows compliance teams to upload any income document — pay stubs, W2s, tax returns, offer letters — and automatically:
- Converts documents to searchable text using real OCR via AWS Bedrock
- Identifies document type automatically
- Extracts key financial fields using keyword matching with confidence scores
- Calculates annual household income across multiple income sources
- Determines AMI eligibility for LIHTC (60%/80% AMI) and MOD (120%/150% AMI) programs based on configurable HUD income tables
- Flags low-confidence extractions for human review
- Learns from human corrections to improve accuracy over time
How we built it
CompliQ is built on a modern full-stack architecture:
- Frontend: Built with v0 and Next.js, deployed on Vercel
- Database: AWS Aurora PostgreSQL for storing applicant records, extracted fields, review results, and the keyword learning library
- AI/OCR: AWS Bedrock with Claude for intelligent document understanding and income extraction
- AMI Engine: Configurable HUD income tables supporting multiple program types (LIHTC TC and MOD) across all household sizes with the 140% Next Available Unit Rule built in
The keyword matching system stores every successful and failed extraction in Aurora PostgreSQL, building a learning library that improves accuracy with every document processed.
Challenges we ran into
The biggest challenge was handling the diversity of income document formats. Pay stubs from different employers use completely different layouts — some say "Gross Pay", others say "Regular Earnings" or "Base Salary". Building a keyword mapping system flexible enough to handle this variation while maintaining high confidence scores required significant iteration.
The second challenge was handling mixed program compliance — a single development can have TC units at 60% AMI, TC units at 80% AMI, and MOD units at 120% and 150% AMI simultaneously. Each unit type has different eligibility rules and the system needed to handle all combinations correctly.
Accomplishments that we're proud of
- Built a document intelligence system that actually understands the nuance of affordable housing compliance — not just a simple calculator
- Implemented the 140% Next Available Unit Rule logic that most compliance software ignores
- Created a self-improving keyword library that gets more accurate with every human correction
- Designed for the real workflow: AI handles the repetitive 80%, humans focus on the 20% that requires judgment
What we learned
Building CompliQ reinforced that the hardest part of AI in regulated industries is not the technology — it is knowing which rules matter and why. The affordable housing compliance domain has layers of federal, state, and local requirements that intersect in ways that only practitioners truly understand.
The combination of domain expertise and technical implementation is what makes this system genuinely useful rather than just technically impressive.
What's next for CompliQ
- Section 8 / Housing Choice Voucher module: automatic HAP contract calculation and NYCHA coordination
- Annual Recertification automation: bulk processing for existing tenants
- TIC Auto-fill: automatically populate Tenant Income Certification forms from extracted data
- Agency Submission Package: generate complete HPD/HDC/EJW submission packages automatically
- B2B SaaS expansion: making CompliQ available to compliance teams across NYC's growing affordable housing pipeline
Log in or sign up for Devpost to join the conversation.