Public sector · Data & AI01
20 years of venue data, unified overnight
Multi-venue public facilities operator
?The problem
Two decades of operational data for a portfolio of major venues sat siloed across disconnected legacy systems. Leadership had no cross-venue view of P&L, budget, or attendance - every report was assembled by hand.
→Our approach
Consolidated the full historical dataset into a governed Microsoft Fabric Lakehouse, built automated ETL pipelines to structure it, then layered Power BI dashboards on top with a data-governance model for long-term auditability.
✓ Reporting time cut from days to minutes
20+ yrsdata consolidated
6+venues unified
Days → minreporting cycle
Transportation · Predictive AI02
Forecasting demand for a regional operator
Regional transportation operator
?The problem
Flight schedules, passenger loads, cargo, fuel, and revenue lived in disconnected systems. Pricing and routing decisions were made reactively from stale summaries, and different teams reported conflicting numbers.
→Our approach
Built a purpose-made predictive analytics platform: unified every data source into one reporting layer, trained ML forecasting models for demand and revenue by route, and shipped dynamic-pricing tools plus real-time executive dashboards.
✓ Reactive reporting replaced with live forecasting
5+sources unified
LiveML forecasting
3stakeholder groups
FinTech · Full-stack SaaS03
One platform, three very different users
Lending platform operator
?The problem
Borrowers, brokers, and administrators each needed a fundamentally different workflow. Without a shared platform, every handoff between them was manual, pricing models were static, and compliance oversight didn't scale.
→Our approach
Architected a single AI-powered SaaS platform with role-specific experiences for all three user types, plus an ML-driven pricing engine for dynamic loan recommendations - built on React, Node.js, Python, and Azure.
✓ Manual handoffs eliminated, one login for every role
3user roles
AI/MLpricing engine
Azurecloud native
Marketplace · Web & mobile04
Zero to production across web, iOS, and Android
Ticket marketplace operator
?The problem
No existing technology foundation, in a price-sensitive market where real-time pricing decisions determine profitability - and no way to track inventory across events and channels in one place.
→Our approach
Delivered a complete full-stack platform from scratch: responsive web app, native iOS and Android apps, an ML pricing-inference engine reading live market signals, and a back-office portal for inventory and reporting.
✓ Live across three platforms in under six months
< 6 moto launch
3platforms shipped
Livepricing engine
Real estate · Web platform05
A search-first property platform, built for one market
Regional real estate technology company
?The problem
Legacy listing tools were fragmented and slow, with no unified way to manage listings, buyer inquiries, and agent workflows - and search wasn't tuned to how local buyers actually look for property.
→Our approach
Built a full-stack web platform with geo-filtered search, an agent portal for listings and lead tracking, and buyer-facing tools for comparison and saved searches - mobile-first, with local market data integrated in.
✓ Agents and buyers moved off manual processes onto one platform
6 moend to end delivery
Mobile-firstdesign
Livemarket data feed
Corporate · Web modernization06
One design system, six outdated sites
Multi-company group portfolio
?The problem
A group of companies under one umbrella ran on outdated, inconsistent websites - different CMS platforms, different code stacks, no shared design system, so every update meant duplicated effort across teams.
→Our approach
Built one shared design system and component library, then rebuilt each company site on a unified modern stack with a CMS non-technical staff could update directly - coordinating delivery across every brand in parallel.
✓ Every site modernized and launched in one coordinated engagement
6 mofull portfolio
1shared design system
Self-serveCMS for staff
Internal · Accessibility07
We scanned our own site and published what broke
KnackHook (internal)
?The problem
Accessibility claims are easy to make and hard to verify, so most vendor statements are unfalsifiable. We wanted ours to be checkable - which meant running our own scanner against our own site and publishing the result whether or not it was flattering.
→Our approach
Scanned all five pages with SureAccess, our Section 508 testing platform. It found 6 landmark failures, and referred 124 contrast conditions it could not decide to human review. We fixed the failures, measured every referred item by hand, corrected one genuinely illegible element the scan had not flagged, then re-scanned and published both reports publicly.
✓ 6 failures to 0, with both scans public
6 → 0confirmed failures
45 → 53rules passed
Publicbefore and after
Wellness · Booking platform08
Replacing phone-and-email booking with self-service
Fitness and yoga studio operator
?The problem
Class booking ran entirely on phone and email - causing double-bookings and heavy administrative load - with no self-service way for members to view schedules, book classes, or manage their accounts.
→Our approach
Built a member-facing booking portal with real-time schedules and instant confirmation, an instructor dashboard for rosters and attendance, a studio admin panel for class and waitlist management, and integrated payments for packs and memberships.
✓ Manual phone and email booking fully replaced with self-service
Self-servemember portal
Liveinstructor dashboard
Automatedpayments & renewals