Projects

Software and data systems for real-estate finance and commerce, from borrower intelligence and deal attribution to pricing and automation.

MoFin · Commercial data infrastructure

Deal Lifecycle & Attribution

Built a centralized view of a deal’s path across data providers, lead-list generation, outreach agencies and sequencers, CRM, communications, and internal processing and underwriting records. Reconstructed historical journeys to connect participants and touchpoints with stage outcomes.

↗ Made fragmented deal histories traceable and usable for evaluation across teams

Data ModelingEntity ResolutionPipedrivePlusVibe / SmartleadSemantic Analysis
  • Historical deal reconstruction at scale
  • Connected email, call, text, CRM, and underwriting records
  • Stage-level signals for engineering, sales, and borrower evaluation
How I built it
  • Linked identifiable people and interactions across services
  • Combined structured records with semantic and statistical analysis where needed
  • Centralized record keeping and visibility into the deal lifecycle
Conceptual flow from source records through reconciliation to financing signals

MoFin · Proprietary analytics platform

Borrower & Lending Intelligence

Built a canonical borrower and loan data platform across vendor, property, CRM and operating records. Field-level provenance, entity resolution and repeatable data rebuilds.

↗ Turned conflicting source records into traceable financing signals

PythonSQLPostgreSQLETLEntity ResolutionWeb Scraping
  • Hundreds of thousands of borrowers processed
  • Corrected over 100,000 loan-status records
  • Transaction-based mortgage reconstruction for borrower targeting
How I built it
  • Resolved related borrower entities
  • Reconstructed mortgage and transaction histories
  • Built models of borrower fit and financing demand
Read case study →

MoFin · Infrastructure and service integrations

Acquisition Infrastructure & Workflows

Built proprietary outbound infrastructure from zero and connected enrichment, public-data processing, CRM records, and campaign delivery through internal codebases and server-side pipelines.

↗ Created an owned foundation for scaling prospect research and outreach

LinuxPostfixCloudflareClayAirtableZapierPipedrive
  • Scaled from zero to hundreds of mailboxes
  • Capacity to contact hundreds of thousands of leads monthly
  • Resumable processing with deduplication and state tracking
How I built it
  • Automated mailbox provisioning and deliverability monitoring
  • Connected custom processing with business-facing tools
  • Turned manual research and handoffs into repeatable workflows
Conceptual flow from authenticated editing through versioned publishing to site content

MoFin · Web development and quality engineering

Lending Website & Visual CMS

Launched the lending website with financial calculators and lead capture, then built a visual CMS for business users. Authenticated in-page editing, versioned publishing, undo history and real-time collaboration.

↗ Gave business users a visual workflow for maintaining site content

Next.jsReactTypeScriptAirtablePlaywrightVitest
  • Financial calculators with testable business logic
  • Validated lead capture connected to Airtable
  • Automated content, visual, and interaction checks
How I built it
  • Translated lending requirements into customer-facing tools
  • Built regression checks for the migration
  • Connected lead intake to the broader operations workflow
Read case study →
LuxeSwap Web Platform screenshot

Full-stack production platform

LuxeSwap Web Platform

Built the company's entire web platform from zero in 6 weeks. The company had no website before this. Full-stack production system with live inventory integration, B2B lead generation, and 8-language internationalization.

↗ Part of the LuxeSwap software suite that contributed to $350K+ year-over-year revenue growth

React 19ViteExpress 5Node.jsRedisSupabaseAWS S3
  • Live inventory integration and B2B acquisition
  • 19 pages with 8-language i18n (incl. RTL Arabic)
  • 30+ qualified B2B inquiries monthly and 500+ cumulative form submissions.
  • Built solo in 6 weeks
How I built it
  • Live eBay Browse API integration with Redis caching
  • "Artisan-Tier" brand showcases with scroll-driven CSS animations
  • Custom Puppeteer scraper for AuctionNinja with cron refresh
  • Dynamic sitemap, JSON-LD structured data, GDPR geo-based consent
View Live →

Internal business platform

Internal CRM & Analytics Dashboard

Built to replace 2,000+ manual spreadsheets. Full internal CRM with JWT-secured API, automated commission calculations, and dual customer segmentation — managing 750+ active consignor accounts.

↗ Replaced 2,000+ manual spreadsheets with unified system

React 19Express 5JWT AuthSupabase/PostgreSQLPython
  • 17 JWT-secured API endpoints
  • 750+ active accounts managed
  • $8.9M+ revenue and 150K+ items processed
  • Python ETL: 2,155 XLSX files → 25 canonical categories
How I built it
  • Dual customer segmentation (volume + value tier)
  • Batch lifecycle tracking with staged-save architecture
  • Automated commission calculations with full edit history
  • Inline editing with undo and audit trail
Brand Pricing Intelligence System screenshot

Data intelligence platform

Brand Pricing Intelligence System

Proprietary brand pricing index built from historical sales data. Cross-category median ratio computation with fuzzy matching and Levenshtein distance for brand deduplication across 1,021 brands.

↗ Powers real-time pricing decisions across 1,021 brands

Node.jsExpressSupabaseReact
  • 130K+ historical sales supporting the pricing engine
  • 1,021 brands classified into 4 tiers
  • 300+ brand aliases normalized
  • 9 analytics visualizations across 14 canonical categories
How I built it
  • Cross-category median ratio computation for tier assignment
  • Fuzzy matching + Levenshtein distance for deduplication
  • Real-time quote generation for consignment triage
  • Powers B2B outreach prioritization
EZ-Lister screenshot

AI-powered desktop application

EZ-Lister

AI-powered desktop app that automates eBay listing creation. Deployed as self-updating macOS app — CEO trained and operates independently in daily production use.

↗ Cut listing preparation from 10 minutes to 30 seconds per item

Pythonsentence-transformersTkinterPyInstaller
  • 95% time reduction (10 min → 30 sec per item)
  • $250K+ annual revenue enabled
How I built it
  • AI classification using sentence-transformer embeddings
  • 77+ categories via word-boundary regex matching
  • Self-updating macOS app with GitHub release detection
  • Packaged a self-updating macOS application for daily business use
Sales Consolidator screenshot

Financial reconciliation tool

Sales Consolidator

Sales reconciliation pipeline matching eBay transaction reports to internal listing records, automating fee redistribution and consignment payout reporting.

PythonTkinterPandas
  • Processes 1,000+ transactions per batch
  • O(n+m) matching algorithm — 60s → <5s execution
How I built it
  • Unified Transaction Dictionary for O(1) lookups
  • Proportional fee redistribution for multi-item orders
  • HTML Meta-Index for batch metadata extraction
  • macOS app bundle for Gatekeeper compliance
Career Intelligence Platform screenshot

University capstone (IBM Watson sponsored)

Career Intelligence Platform

Led cross-functional team sponsored by IBM Watson researcher (Dr. Chidansh Bhatt). RAG-based career matching engine with semantic job-resume matching and geospatial salary visualizations.

PythonLangChainOpenAIMongoDBPlotlyMapbox
  • RAG-based career matching with OpenAI embeddings
  • 50-state salary choropleth visualizations
  • CareerOneStop (Dept. of Labor) API integration
How I built it
  • Semantic job-resume matching via OpenAI embeddings
  • GPT-powered resume and job listing parsers
  • Plotly choropleth salary maps across 50 states
  • Mapbox geographic analytics for applicant data