ML/AI Engineer, Data Scientist, Economist

Portrait of Maximillian Mantei

Profile

Bridges machine learning, agentic AI, data engineering, and MLOps into scalable end-to-end solutions. Combines rigorous statistical foundations and causal inference with economic acumen to develop robust, value-driven data products and sustainably enable data-driven decision-making.

Tools & Expertise

  • ML Engineering

    • PyTorch
    • polars
    • numpy
    • scikit-learn
    • tidyverse
  • Agentic AI

    • LangChain
    • LangGraph
    • fastmcp
    • Pydantic AI
  • Python

    • asyncio
    • pytest
    • uv
    • ruff
    • mypy
    • FastAPI
    • SQLAlchemy
  • MLOps & LLMOps

    • MLFlow
    • Docker
    • DVC
    • git
    • GitHub
    • AWS
    • Linux
    • HuggingFace
  • Data Engineering

    • Airflow
    • prefect
    • duckdb
    • InfluxDB
    • Postgres
    • MQTT
    • Kafka
  • Data Frontend

    • Streamlit
    • plotly
    • Grafana
    • NiceGUI
    • Vega-Altair
    • shiny
    • ggplot2
  • Statistics & Econometrics

    • Bayesian Optim.
    • Causal Inference
    • Gaussian Processes
    • Timeseries

Certifications

Languages

  • GermanNative language
  • English4/5
  • Spanish2/5

Experience 7 entries

Freelancer

welded.io UG

Mettmann, Germany (remote)

  • Owned the design and architecture of a scalable data platform for processing high-resolution IIoT real-time data
  • Implemented an AI-based computer vision pipeline end to end for visual monitoring of industrial processes

agital.online GmbH

Lüneburg, Germany (remote)

  • Designed and implemented a large-scale, continuous monitoring system for heating oil prices
  • Developed and deployed an interactive dashboard for data-driven market monitoring
  • Built a data lake enabling high-performance analysis of over 2 billion price data points

Nahnetz GmbH

Hamburg, Germany (remote)

  • Responsible for the design, modeling, and evaluation of market and preference analyses (including conjoint analyses), leveraging Bayesian models to deliver robust market potential estimates
  • Developed simulation-based decision-making models for assessing and optimizing the economic viability of battery storage systems, leveraging probabilistic methods (Gaussian process regression)

Raiffeisen Networld GmbH

Cologne, Germany (remote)

  • Developed predictive vision models to estimate soil organic carbon content by integrating satellite imagery and in-situ data
  • Developed probabilistic demand forecasts to optimize data-driven B2B customer campaigns, incorporating uncertainty quantification to support more robust decision-making

Causiq AB

Stockholm, Sweden (remote)

  • Owned the design and implementation of a tailored Bayesian media mix model for measuring the causal impact of marketing channels
  • Developed an attribution model based on Shapley values for fairly attributing marketing effects

Research Assistant and Tutor

  • Contributed to DFG-funded research projects in behavioral economics and empirical economic research, including designing and evaluating randomized controlled trials (RCTs); built and analyzed complex panel datasets using econometric models; taught statistics and programming (R) as a lecturer and tutor.

Education 5 entries

PhD Candidate in Economics

Universität Hamburg

Hamburg, Germany

DFG Research Training Group: Internationalization of the Law and its Economic Analysis · DFG full scholarship

  • Bayesian semiparametric hierarchical GLM estimation of foreign direct investment flows and stocks based on dyadic panel data
  • Developed causal models to account for endogenous sample selection and non-random missing data in order to identify heterogeneous contract effects

MSc Politics, Economics, and Philosophy

Universität Hamburg

Hamburg, Germany

  • Hierarchical Bayesian analysis to model the behavioral-economic effects of individual bidding behavior in Dutch auctions (reverse auctions)

Erasmus Exchange

Aarhus Universitet

Aarhus, Denmark

B.A. in Political Science and Economics

Universität Bremen

Bremen, Germany

  • Quantifying strategic voting behavior in European interwar democracies through regression analyses

Abitur (German university entrance qualification)

Freie Evangelische Bekenntnisschule Bremen

Bremen, Germany