Hi, I'm Andre.
Senior Backend Engineer
I like building things that work reliably and explaining why they work that way. When I'm not with my family, I'm probably deep in a board game or caught vibe-coding after everyone's gone to sleep.
I've spent the last 12+ years in Java backend, data infrastructure, and cloud platforms. Most recently I built and scaled Joblift's revenue-critical backend (ingestion, pricing, and search) over six years in a high-throughput microservice architecture on Google Cloud.
Experience
- Partnered with one other engineer to redesign 10+ microservices handling 100M+ daily job updates across six markets in Europe and the US. Replaced a MySQL deduplication bottleneck with Bigtable, shifted Kafka from heavy payloads to lightweight meta-messages backed by Bigtable lookups, and switched from sequential to parallel microservice processing, which cut average time-to-index from over an hour to 10–15 minutes. That also reduced revenue leakage from expired clicks and improved gross margin.
- Built the Spring Boot microservice that became the default pricing option for Joblift's flexible-pricing customer segment. Integrated with Joblift's Elasticsearch-backed search service to continuously adjust per-click pricing for 100k+ jobs by weighing impressions, clicks, and ranking position against each job's remaining customer budget, which paced spend more evenly over a job's lifetime and improved budget utilization. In production for over two years without a single outage.
- Took on a growing share of 100+ microservices as the engineering team shrank from around 15 engineers to 2, and kept production systems stable through restructuring.
- Designed and built "Dataheart" — a Hortonworks HDP Hadoop/HBase platform on private cloud to support the company's international expansion. Replaced the legacy data warehouse's one-way batch model with near-real-time ingestion and reverse flows of aggregated data back into production systems, which enabled product-side personalization and marketing feedback loops.
"When I have a problem I'm always happy to ask you because I know I will not just get a solution to my problem but I will actually understand much better what my real problem is."— Jan Stolle, peer review 2018
- Operated and extended the HP Vertica + Cloudera Hadoop analytics platform processing hundreds of millions of daily tracking events across the company's online and mobile games. Joined to support the zero-downtime migration from the legacy MySQL data warehouse to the 5-node Vertica cluster.
- Scaled Vertica from 5 to 7 nodes and the Cloudera Hadoop cluster from 6 to 8 nodes, supporting hundreds of TB of raw data and 100+ ETL jobs serving marketing, game, and product analytics across the company. Used Ansible-based provisioning with the IT operations team for the cluster extensions, and integrated HBase into the platform.
- Designed and delivered company-wide Vertica best-practices training for analysts, improving query quality and reducing the impact of poorly written ad-hoc queries on cluster stability.
- Built a Java + JSoup pricing tool that crawled competitor prices on B2B tire-sales platforms and generated price recommendations through a custom backend and frontend.
- Extended EvoSuite, an open-source Java test-generation research project, via a bachelor's thesis and subsequent Werkstudent work on bytecode analysis and instrumentation.
- Developed leginda.de in PHP for a German translation agency, building their order portal over four years.
- Tutored the introductory programming course at the Chair for Dependable Systems and Software (Prof. Hermanns).
Tech Stack
Education
Universität des Saarlandes, Saarbrücken
Grade: 2.0
"Search Based Dataflow Testing"
Grade: 1.0
Projects
Interests
Modern, heavy strategy games — the kind where the rules explanation takes an hour and you love every minute of it.
Building things for fun with AI as a copilot. Lothar started as one of these sessions.