Maximilian Kretzschmar

MAN Truck & Bus SE

• Architecting AMPLIFY, a multi-agent system orchestrating 7 specialised agents with 40+ tools automating FE simulation workflows in a human-in-the-loop system for ~50 engineers across Germany and India; MAN IT Innovation Award finalist, projected to save ~10% of engineering hours across key simulation workflows
• Built React/TypeScript frontend for real-time agent monitoring with execution graphs, 3D visualisation, human feedback and test & evaluation tooling; validated with 7 domain experts using a 23-item assessment framework co-developed with University of Bayreuth
• Developed Ansys DPF tool integration layer as part of the modular multi-agent framework, connecting simulation data via gRPC across 3 analysis domains for production vehicle models
• Deployed on AWS (EC2, Bedrock, S3, Cognito) with ~€15,000 infrastructure budget, enabling scalable execution of workflows and reducing manual intervention in engineering analysis pipelines
• Leading €320,000, 3-year industry research project as sole researcher managing stakeholders, aligning academic goals with MAN's AI development roadmap including project scoping, legal coordination, budget management, delegation of research activities
• Leading the cross-disciplinary development curriculum for 30+ PhD candidates organising multi-day workshops
• Managed end-to-end recruitment, screened 70+ candidates for student research positions, supervised several master's students across 6-month placements

Dresden University of Technology

• Researching hybrid AI systems that combine LLMs with structured and unstructured knowledge representations to automate product development processes
• Designing and evaluating multi-agent architectures with RAG pipelines and knowledge graph construction for engineering data retrieval
• Establishing and managing research partnerships with academic and industry partners, including project scoping, legal coordination, finances and delegation of research activities
• Authored and presented 9 peer-reviewed papers at international journals & conferences (Cambridge University Press, IEEE, Elsevier)
• Coordinating 8-person team delivering coursework to ~250 students/semester
• Teaching tutorials for groups of 30 students and supervising several bachelor's and master's theses

Bosch

• Implemented the Bosch Center for AI's Anomaly Detection Engine across 9 fully automated drilling laser lines producing ~26,000 parts per day/€9.5M annual production value; identified systematic quality anomaly causing redundant testing, enabling rapid correction of throughput losses
• Built Python tooling (pandas, NumPy, seaborn) for automated preparation, analysis, and visualisation of machine and CMM data across daily production cycles
• Planned, led and moderated a factory layout workshop with a 15-person cross-functional team affecting 25+ workers in manufacturing, driving process improvements using lean production methods such as one-piece-flow
• Independently identified a recurring assembly process inefficiency, prototyped a 3D-printed solution, documented for replication; adopted as standard across plants in 3 countries (Germany, US, Turkey)

Fraunhofer IWS

• Designed, developed, and integrated hardware for hyperspectral imaging systems, including the institute's smallest-ever diffuser component, from concept to manufacturing-ready documents, coordinating 3 teams

SPEZ GmbH

Dresden University of Technology, Germany

Dresden University of Technology, Germany

1,5 · First Class Honours

European School Culham, United Kingdom

87.21% · 1,3 · AAA

Career Strategy · Impact Assessment · Cause Prioritisation · Theory of Change

Centre for Effective Altruism

Stakeholder & Environment analysis · Phase Plan & Milestones · Agile Project Management

TUD Graduiertenakademie

Value Chain · Procurement & Operations · Sales & Marketing · Accounting · Controlling

TUD Graduiertenakademie

Stakeholder Analysis · Communication Strategy · Conflict Resolution

CapGemini

Research Methodology · Literature Review · Data Collection & Analysis · Evaluation Methods

The Design Society

AI Lifecycle · Transformers · Fine-Tuning · Scaling Laws · Model Evaluation · Deployment

DeepLearning.AI · Amazon Web Services

Algorithms · Data Structures · Software Engineering · C · Python · SQL · JavaScript

Harvard University

Data Structures · Web Data · Databases · Data Visualisation

University of Michigan

Personal website & private LLM console

• Built a full-stack portfolio application using Next.js 14, Supabase, and the Vercel AI SDK with production deployment on Vercel
• Developed an authenticated LLM console with multi-provider inference, real-time streaming, and session-based interaction

3-axis CNC plasma cutter

• Designed and built a 3-axis CNC plasma cutting system from scratch -- mechanical design, electrical integration, software, sourcing, and testing -- to launch metalworking business for under £1,000

Object detection with CNNs trained on synthetic data

• Investigated domain gap effects on CNN models using synthetic data for object detection of real machine components via custom Blender pipeline & trained YOLOv7 model

The School for Moral Ambition

• Founded international Moral Ambition Circle recruiting 6 professionals across 5 countries spanning AI research, digital policy, and management to explore pathways to greater positive impact
• Facilitating the 8-month peer-led programme where members develop personal Theories of Change connecting their skills to measurable impact on pressing global challenges

Kait-Bu-Rin Muay Thai

• Volunteer martial arts instructor for groups of children and teens

Kuzuko Foundation South Africa

• Converted a disused barn into an environmental education centre and organised fundraising events to support the initiative

Human-AI Interaction · Affordance Theory · Transparency Evaluation Concept

International Symposium on Hybrid Intelligence in Product and Production Engineering

Knowledge Graphs · Schema Creation · Neo4j

Digital Chemical Engineering

Knowledge Graphs · Agentic Schema Creation · Neo4j

Advances in Artificial Intelligence & Machine Learning (AAIML 2026)

Compound LLM Systems · Industry Case Study · Systems Design Patterns

Elsevier

Industry Survey · Literature Review · Responsibility Model

Cambridge University Press

MAS Development Framework · Workflow Automation · Product Development

IEEE

Systematic Literature Review · Cross-Domain Analysis

The Design Society

Capability-Challenge Matrix

TU Dresden

Synthetic Data Generation · CAD-to-training Pipeline · Domain Randomization

Cambridge University Press