Lars V.
✓ Vetted CrafterPrincipal Software Engineer – Performance & Deployment
I optimise performance and deploy reliably — scale and reliability are my focus.
About
I optimise performance and deploy reliably. I have 15 years building systems at scale — from Booking.com to Philips. I specialise in performance optimisation, deployment pipelines, and scalable architectures. Cursor helps me write IaC and refactor deployment scripts. I work with TypeScript, Node.js, Docker, and Kubernetes.
My focus is on systems that scale and deploy safely. I have reduced API latency by 70%, built zero-downtime deployment pipelines, and optimised database queries for high-traffic workloads. I care about observability: metrics, tracing, and dashboards that tell you when something is wrong before users notice.
AI Expertise
Notable Projects
Petabyte-Scale Lakehouse Migration
Led architectural design and execution of migrating a Dutch media company from a Teradata + Hadoop stack to a Delta Lake lakehouse on Azure. Redesigned 400+ pipelines using Apache Spark, implemented dbt for transformation layers, and built a real-time Kafka integration for event streaming.
✓ Annual infrastructure cost reduced by $2.1M; query performance improved 5-10x for analytical workloads; ML training job times reduced by 60% with Delta Lake optimized reads.
Real-Time E-Commerce Recommendation Platform
Designed and built a two-tower recommendation system with real-time feature serving for a Benelux e-commerce platform. Offline training on Spark using historical purchase and browse events, online feature serving from Redis with 30-minute incremental updates via Kafka Streams, and a FAISS-based ANN retrieval layer.
✓ Recommendation click-through rate improved 24% vs. rule-based system; revenue attributable to recommendations increased from 18% to 29% of total GMV.
Streaming Feature Store with Feast
Built a production feature store using Feast with dual-store architecture: Kafka Streams for online feature computation, Redis for low-latency serving, and Delta Lake on Azure ADLS for offline training. Implemented a custom Feast materialization job for streaming features.
✓ Feature serving P99 latency of 8ms; 200+ features shared across 15 ML models; eliminated duplicate feature engineering code that previously existed in 12 different team repositories.
Work Experience
Principal Data Platform Engineer
Booking.com
2017 – Present
Define the data platform architecture strategy for 1.5 million daily active users worth of analytical and ML workloads. Lead a team of 12 engineers across data infrastructure, feature engineering, and ML platform.
Senior Data Engineer
Philips
2010 – 2017
Built data infrastructure for healthcare analytics and medical device telemetry. Introduced Hadoop-based processing to replace overnight batch jobs that were missing SLAs.
Education & Certifications
M.Sc. Computer Science (Data Management)
Delft University of Technology · 2009
Interested in working with Lars?
Tell us about your project and we'll facilitate an introduction.