Three minutes to five seconds. One country to many.

Rebuilding a rental housing alert platform for speed and international expansion, while it kept growing.

ClientStekkies

IndustryPropTech

EngagementCore engineering team

TimelineOngoing, since 2023

Websitewww.stekkies.com/en

Who they are

Stekkies is a fast-growing housing alert platform that helps renters find properties before anyone else does. As the company expanded internationally, it needed an architecture that could support multiple countries, more listing sources, AI-powered data processing, and genuinely real-time notifications. Enable international expansion, cut alert delivery time, widen housing coverage, automate property data extraction with AI, and leave behind an architecture flexible enough for whatever market came next.

What the work moved.

60×

Faster alert delivery: 3 minutes down to 5 seconds

Average time from listing published to alert in a user's hand

2 markets

New countries launched on the refactored architecture: UK and Germany

2023

Embedded as a core engineering team member since

What we owned.

Ignicube worked as part of the Stekkies team, not a vendor on the other side of a ticket queue.

Multi-country refactor & configuration architecture
Real-time alert optimization (Celery queue workflows)
Adapter pattern implementation for service providers
Scalable data source expansion
AI-assisted scraper development
AI-powered property information extraction

The challenge

What stood
in the way.

Single-country architecture

The platform was built with hardcoded logic for one country, making every new market a rewrite rather than a configuration change.

Three-minute alert delay

Property alerts took roughly three minutes to reach users. An eternity in a rental market where the first responder gets the viewing.

Tightly coupled providers

Service providers were wired directly into application logic, so swapping one or adding a country-specific alternative meant touching code everywhere.

Limited listing coverage

Staying competitive required more housing sources, and each new one was a bespoke integration effort.

Information buried in free text

Rental requirements, amenities, pricing details, and special conditions all lived inside unstructured descriptions, invisible to search and filters.

What we built

How we
solved it.

Multi-country refactor

Removed hardcoded single-country logic across the codebase and introduced configurable wrappers around key functionality, enabling country- and company-specific configuration as data rather than code.

Real-time alert pipeline

Redesigned alert generation to eliminate redundant computation and split one compute-heavy flow into several efficient stages, taking average delivery from three minutes to five seconds.

Adapter pattern for integrations

An adapter-based architecture for service providers, so integrating, replacing, or testing a provider in a new country no longer touches application logic.

Scalable source ingestion

A framework for onboarding new housing listing sources consistently across countries, expanding market coverage without multiplying maintenance.

AI-assisted scraper development

Used AI to accelerate building and maintaining scrapers, cutting the effort to add a source and to keep up as target sites change.

AI information extraction

LLM-based extraction that structures the details hidden in free-text listings, powering richer filters and far more accurate property matching.

Impact

What changed for Stekkies.

Alert delivery time dropped from roughly 3 minutes to 5 seconds, a 60× improvement.
Successfully launched in the United Kingdom and Germany on the new multi-country architecture.
Plug-and-play provider architecture made onboarding and testing new integrations routine.
Additional housing data sources integrated, materially increasing listing coverage.
AI-assisted development cut the time to build and maintain new property scrapers.
Structured extraction from listing descriptions enabled richer search filters and better matching.

Technology

PythonDjangoCeleryPostgreSQLScrapyOpenAILLMsAdapter Pattern

Disciplines

PropTechData PipelinesAI DevelopmentAI Agent Development

The Ignicube team functioned as core members of our own team. We've had incredible growth as a company, and they were able to handle all challenges thrown at them with ease and autonomy. Primary achievements consisted of optimizing Celery queue workflows, introducing factory patterns for key solution providers and rewriting our codebase to handle multi-country configurations.

Founder · Stekkies

Keep readingAll case studies