Head of Data Platforms & Analytics + Interim Lead-to-Cash (L2C) Lead — Digital Charging Solutions (DCS)
e-mobility JV owned by BMW, Mercedes-Benz & BP · Berlin · interim L2C lead since Jan 2026
Mar 2025 — Present
- Voucher self-service tool (product + business process owner): turned an ownerless, fragmented 7-step process into a single automated page covering product → pricing → contract → invoice → voucher creation; the cleanup + tool cut uninvoiced promotional vouchers from €40M to €4M (~90%) over two years and surfaced €2M of expired, un-invoiceable losses. Built in Java/Kotlin, integrated the SAP API (create invoices, read GL accounts), with Keycloak auth.
- Led the data discovery and documentation supporting accounting and EY through a tax overhaul and SAP consolidation; cleared ~€100M of transaction credit/liability accounts and reclassified tax on services with the tax team.
- Built an AI model joining contracts, charging activity, vouchers and invoicing that flags activity drops and customer-behaviour shifts after pricing/package/offer changes, to read customers better and steer commercial decisions.
- Built and productionised a machine-learning forecasting model and the long-term plan (LTP).
- Cut Snowflake spend from €160k to €120k/year (~25%) on takeover while the data estate grew to 66 TB.
- Negotiated contracts, built monetisation plans and cash-flow-positive structures, and ran invoice checks; delivered strategic presentations to shareholders, in direct contact with BMW, Mercedes-Benz and Kia HQs.
- Runs ELT/ETL pipelines (dbt, Snowflake, GCP Dataflow, Python, Postgres) and Agile delivery — Scrum sprints on a JIRA board for build, Kanban for operations — leading engineering, analyst and operational teams as a hybrid of data and product.
- Defined a 2-year data strategy aligned to business demand, consolidating and optimising all dashboards into a single source of truth.
- On the data side we build the long-term plan (LTP) with data-science models and a read on customer behaviour: the plan comes out of models of how customers actually behave, not out of a spreadsheet extrapolation.
- Continuously enhancing the data ingestion and semantic layers so the platform is AI-native: a model or an agent queries the same governed layer the business reports from.
- Real-time data through Event Hub feeds the platform alongside the batch pipelines.
- Day-to-day stack: Snowflake as the warehouse, Python and SQL for the work, Power BI and Streamlit for what people see, Event Hub for the real-time feeds.
- Own strategy planning and the long-term plan for the data organisation: where the platform goes, what it costs, and what it has to serve next.
- The platform serves five domains end to end: charging detail records, customer behaviour, internal processes, contract-to-cash finance, and voucher and promotion campaigns.
- Lead the data organisation across Europe and Japan at DCS for several years, after a global demand-planning role at Tesla coordinating Berlin, Shanghai, Fremont and Texas — regional analytics leadership in a multi-market, cross-timezone setup.
- Own the KPI framework at DCS: KPI definitions are set centrally, the business is aligned on them, and reporting standards and governance follow from that single definition.
- Run the platform as Data-as-a-Service: build the semantic layers, onboard business teams onto them, and train stakeholders on the dashboards and self-service analytics so decisions are taken on the data, not around it.
- Automated the Lead-to-Cash monitoring on receivables: the open-receivables position is checked automatically instead of pulled by hand, so a stuck balance surfaces by itself rather than at month-end.
- Automated OEM and fleet invoice monitoring: invoicing to the OEM and fleet partners is checked continuously rather than sampled, so errors and gaps are caught before the customer sees them.
- Developed and automated the bad-debt reporting: what was a manual assembly is now a standing report, so finance and the commercial teams read the same numbers from the same place.
- Transforming the whole dashboard estate: the AI-generated HTML dashboards are migrating onto Streamlit, alongside Power BI, on Snowflake — one place the company reads its numbers from instead of a per-team patchwork.
- We build and automate internal websites, so an automation ships as a tool people actually use rather than as a process document — the voucher self-service page was the first of them.
- The team grows with the demand and now serves the whole company: data that used to answer one department's questions is a shared service across DCS.
