Moritz Hoferer

Data enthusiast & cyclist

Moritz Hoferer

Data analyst working across analytics engineering, data analysis, machine learning experimentation, and full-stack prototyping.

This site is currently a work in progress, but it will eventually serve as a portfolio of my work and projects.

View ProjectsContact

Base

Zurich, Switzerland

Focus

Data Engineering, Analytics & Science

Stack

SQL, Python, Looker, Dataform

Mode

Customer and product analytics

Focus Areas

Work shaped around analysis that can be used.

The structure is intentionally narrow: build stable data foundations, answer concrete business questions, test where modeling helps, and ship small interfaces when they reduce friction.

Analytics & Data Engineering
Build clean analytical models, define reliable metrics, and shape data flows that stay usable over time.
SQLBigQueryDataformLooker
Machine Learning & Modeling
Prototype supervised models, frame experiments, and evaluate where modeling adds actual leverage instead of noise.
PyTorchscikit-learnTensorFlowBigQuery ML
Product / Event Data Analysis
Analyse customer behavior, subscriptions, and event streams to improve retention, reporting, and decision quality.
TrackingRetentionE-commerceControlling
Full-stack Prototyping
Turn analytical ideas into small internal tools and frontends that are fast to test and useful in practice.
Next.jsTypeScriptFlaskAPIs

Selected Work

A compact set of analytical problems worth showing.

These cards use placeholder project framing for now, but the format is final: problem, stack, and a short statement about what the work changed or clarified.

Subscription Cohort Model
Created a reusable view of subscription behavior to compare acquisition cohorts, churn windows, and revenue retention.
BigQuerySQLLooker

Made cohort reporting comparable across teams and exposed where retention moved independently from top-line growth.

Outcome / learning
Event Tracking QA Layer
Mapped a noisy event taxonomy into a smaller analytical layer that could support product questions without manual cleanup.
DataformTypeScriptLooker Studio

Reduced ambiguity in product metrics and made new dashboard work faster because core events became stable inputs.

Outcome / learning
ML Experiment Sandbox
Tested a lightweight classification workflow for campaign or customer-response questions with emphasis on baseline quality.
PythonPyTorchscikit-learn

Clarified when model complexity helped and when a simpler analytical baseline was the stronger operational choice.

Outcome / learning

Visual Insights

Small charts, focused signal.

The chart layer is intentionally quiet. It uses a single accent family and placeholder values to show the shape of the future analytical storytelling without slipping into dashboard theater.

Skills by domain
Relative emphasis across the main working areas.
Analytics engineering92
Data analysis88
Machine learning72
Full-stack prototyping68
Project timeline
Illustrative momentum across recent years.
2021
2022
2023
2024
2025
Work distribution
Indicative split across recurring work modes.
Analytics
42%
Engineering
24%
ML
18%
Frontend
16%

Experience Snapshot

Responsibilities and themes rather than a full resume dump.

The section is deliberately compressed. It highlights the environments, questions, and methods that define the work, without trying to reproduce a full CV inside the landing page.

Now

Data Analyst

Consumer product and subscription environment

Work across reporting, analytical modeling, experimentation support, and the data structures behind recurring business questions.

Customer analysisMetric designDecision support

Earlier

Research and modeling

Political dynamics and complex systems

Applied quantitative modeling to opinion formation, campaign behavior, and structured empirical questions.

InferenceSimulationNetwork thinking

Foundation

Physics training

Munich and Milan

Built a strong quantitative base through theoretical and computational work, later transferred into applied analytics.

Mathematical modelingScientific computingResearch design

Tech Stack

Grouped by where each tool tends to matter.

Instead of long skill inventories or progress bars, the stack is organized by working context: data, modeling, frontend, backend, and the supporting tooling around them.

Data

SQLBigQueryPythonDataformLookerLooker Studio

ML

PyTorchscikit-learnTensorFlowBigQuery ML

Frontend

ReactNext.jsTypeScriptTailwind CSS

Backend

FlaskAPIsData models

Tools

GitBunGCSRecharts

Notes

Writing previews, kept short.

The final notes section will stay factual and narrowly scoped. For now, these cards reserve the structure without pretending the content is finished.

Metric layers that survive product change
Placeholder note on keeping analytical models stable while event schemas and stakeholder questions keep shifting.

Placeholder preview

When baseline models are enough
Placeholder note on choosing simpler ML baselines before investing in heavier experimentation or model operations.

Placeholder preview

Useful dashboards are narrow dashboards
Placeholder note on reducing reporting surfaces to the few views that actually support a recurring decision.

Placeholder preview