redecvi

• Led the modernization of the Web3 data platform, building cloud-native pipelines ingesting multi-gigabyte on- and off-chain Ethereum datasets including smart contract events and validator metrics, reducing runtime and costs by ~70%. • Migrated batch pipelines to real-time streaming, shrinking data-freshness lag from 24 h to under 1 min and enabling time-sensitive DeFi analytics products. • Led end-to-end development of a rewards and incentives engine processing millions of on-chain events daily, adopted by 100% of clients and running bug-free in production. • Built Grafana and Dune dashboards exposing validator health, slashing-risk events, and DeFi protocol anomalies to engineering, product, and client teams in real time. • Introduced dbt and ClickHouse with automated data quality tests in CI/CD, and mentored distributed team members on DataOps best practices. • Tested Ethereum validators and nodes, including Obol distributed validator clusters, and contributed Solidity smart contract code to one project.

• Built an internal LLM-powered RAG system with LangChain, Llama 2, Qdrant, and a Neo4j knowledge graph, enabling natural language queries over internal documentation. • Owned the full ML lifecycle of production ML infrastructure in Python (PyTorch, MLflow): automated training, evaluation, deployment, drift monitoring, and CI/CD. • Engineered features for a fraud and anomaly detection system on time-series transactional data, switching from Random Forest to Gradient Boosting to reduce false positives by 30%. • Refactored a legacy application through algorithm optimization and parallel processing, cutting runtime 97% (3 h to 5 min) and adding documentation and unit tests.

• Built production Python backend applications automating internal processes and integrating REST APIs with legacy SQL databases, saving ~5 FTE per year and eliminating manual data entry. • Developed a document metadata extraction and summarization pipeline for scanned PDFs using OCR and NLP, extracting structured information from thousands of documents monthly; adopted by Operations teams across Switzerland. • Led cross-functional development of a scalable capacity-planning application, ingesting 20+ data sources and forecasting workload from time-series data; deployed across Europe.

• Developed a Python / PostgreSQL KPI tracking tool for a Swiss bank, replacing manual Excel processes across ~15 teams, and a project cost-reporting tool saving ~1 FTE. • Collaborated on the IT Target Operating Model for a Finnish bank, leading to a €200 m implementation deal.