The Challenge
joblocal operates one of the leading platforms for regional job listings in Germany. The core business: generating reach for job postings and targeting them on relevant channels. However, the existing AdTech stack was based on a legacy PHP system that could no longer meet the growing demands for performance, scalability, and real-time processing.
At the same time, joblocal didn’t just want a new system - they also wanted to enable their own development team to maintain and extend it long-term. The challenge was therefore twofold: technical modernization and sustainable skill building.
The Approach: Pair Programming
Instead of a traditional contract project, we chose a collaborative approach: innFactory developers worked for several months in mixed teams directly with joblocal developers. Every line of code was written together, every architecture decision was made together.
This approach brings several advantages:
- Immediate Code Review: Errors are caught during development, not weeks later
- Context Transfer: The joblocal team knows every corner of the new system because they were actively involved in its development
- Sustainable Enablement: After project completion, the team can continue development independently
The Technical Solution
Targeted Performance Service
The new system enables intelligent campaign management for job listings. Based on data analysis, ads are automatically placed on channels that offer the highest conversion probability for the respective target audience.
Modern Architecture
| Component | Technology |
|---|---|
| Backend | Play Framework, Scala |
| Data Lake | Databricks, Apache Spark |
| Streaming | AWS Kinesis |
| Database | DynamoDB |
| Analytics | BigQuery |
| Cloud | Multi-Cloud (AWS & GCP) |
Data Lake with Databricks
The actual intelligence of the system lies in the data lake: campaign performance data, click rates, and conversion metrics flow together here. Spark jobs analyze this data and provide the foundation for automated campaign optimization. With Databricks, data scientists and developers can collaborate on ML models without worrying about infrastructure.
Multi-Cloud Strategy
The system deliberately uses services from both major cloud providers: AWS Kinesis for real-time streaming, DynamoDB for fast key-value access, but BigQuery on GCP for complex analytical queries. This best-of-breed strategy enables optimal performance at reasonable costs.
The Results
After several months of intensive collaboration, joblocal now has a modern, scalable AdTech stack that meets the demands of real-time campaign management. More importantly: their own development team has built the competence to independently extend the system and implement new features.
The project demonstrates how technology transfer can work: not through documentation or training, but through working together on the real system.
Related Services: Software Development | AI & Machine Learning
