We inherit broken codebases, revive them, then modernize under real pressure. Dwellago runs across four phases — revival, migration, interactive matching, and live AWS plus store redeployment.
01Phase 1·US real estate lead-generation platform
Dwellago
Codebase Revival
We inherited a broken stakeholder platform — unstable, hard to ship, and no longer trusted by the team around it. Before any migration talk, we revived the codebase and made it operable again.
Challenge
The system had been left in a failing state. Builds were fragile, ownership was unclear, and product progress had stalled. Nobody could migrate what they could not safely run.
What we did
We took ownership of the inherited codebase. Stabilized critical paths, restored confidence in builds and releases, cleaned the worst failure modes, and got the platform back to a place where engineering work could compound.
Outcomes
Inherited and stabilized a broken production codebase
Restored release confidence for the stakeholder team
Made the platform operable again before modernization
With a living codebase back in hand, we migrated the platform onto current standards — framework modernization, enterprise schema overhauls, and a serious test suite.
Challenge
The revived platform still sat on outdated frameworks and schemas. It needed Flutter and Golang upgrades, database redesign for better categorization and utility, and test coverage that could support real product work.
What we did
We owned the migration. Framework modernization across frontend and backend, schema redesign for production load, and unit testing toward high coverage so later product and infrastructure work had a stable base.
Outcomes
Framework optimization across Flutter and Golang
Enterprise schema overhauls completed
Comprehensive testing coverage in place
Platform ready for product and infrastructure phases
FlutterGolangPythonDatabase DesignTestingAPI Development
03Phase 3·US real estate lead-generation platform
Dwellago
Interactive Matching & Lead Generation
We built a matching algorithm that works against a dynamic system of questions, answers, and tags — so sellers and buyers can generate high-quality leads through an interactive flow.
Challenge
Static listing flows were not enough. The product needed buyers and sellers matched through evolving preferences — questions, answers, and tags that change as users engage — without collapsing into low-quality noise.
What we did
We designed and shipped a matching layer over the dynamic Q&A and tagging system. Properties and users are scored against live preference signals so lead generation stays interactive and high-signal for both sides of the market.
Outcomes
Matching algorithm over dynamic questions, answers, and tags
Interactive buyer–seller lead generation
Higher-quality property matches under real preference drift
Product layer built on the revived and migrated stack
04Phase 4 · In progress·US real estate lead-generation platform
Dwellago
AWS Infrastructure & Store Redeployment
Current work: operating the complex backend on AWS (including ECS and related services), and getting the apps back into the Play Store and App Store after older pre-migration builds were pulled.
Challenge
The modernized backend needed production-grade AWS management — container orchestration, service topology, and operational discipline. In parallel, store listings were down because old pre-migration app versions had been pulled, blocking distribution.
What we did
We are managing AWS deployment for the complex backend — ECS and the surrounding infrastructure — while preparing and executing Play Store and App Store redeployments so post-migration builds replace the pulled versions and the product is shippable again.
Outcomes
AWS backend operations underway, including ECS
Infrastructure ownership for a complex production backend
Play Store and App Store redeployment in flight
Replacing pulled pre-migration builds with current releases
Custom OCR pipeline that turned a decade of paper legal archives into a searchable digital system.
Challenge
Ten years of paper records. Variable scan quality. Multiple formats. Staff were burning hours on retrieval.
What we did
We built an OCR pipeline for legal documents, extracted text across formats, and made the archive searchable so teams could find records without digging through boxes.
Outcomes
Digitized a 10-year legal archive
End-to-end searchable document set
Faster retrieval for legal staff
Historical records preserved digitally
PythonOCRDocument ProcessingText ExtractionSearch Systems
Inherited a broken codebase?
Tell us what you took over. We will tell you what is worth stabilizing first.