Ralph Drott

// ML Engineer · Edge AI · Autonomous Systems

Hey, I'm Ralph.

CS student specializing in AI and Machine Learning, transitioning from five years of industry research at a Fortune 500 company into ML engineering.

At Procter & Gamble I built dashboards, ran statistical analyses, and led consumer studies. Now I'm applying that data mindset to Edge AI, building models that run on constrained hardware, not just in the cloud.

Based in Trebur, Germany · Open to ML Engineer roles in the Rhine-Main area.

View Resume See Projects Download CV

// 01 · Resume

Background & Skills

Experience

2025 — 2026

Researcher

Procter & Gamble Service GmbH · Schwalbach am Taunus

Led consumer studies and projects for new product initiatives in a production environment. Continued responsibilities from the Associate Researcher role with expanded project ownership.

2021 — 2024

Associate Researcher

Procter & Gamble Service GmbH · Schwalbach am Taunus

Statistical data analysis (significance testing with JMP), consumer study evaluation, PowerBI dashboard development, lab reporting in English, and development of standardized test methods.

2020 — 2021

Physics Laboratory Technician

Primes GmbH · Pfungstadt

Operation and calibration of high-power laser measurement systems. Detailed technical documentation, with expertise in mechanics, optics, and sensor technology.

Education

2023 — Present

B.Sc. AI & Machine Learning

Wilhelm Büchner Hochschule · Darmstadt

Started part-time alongside industry work at P&G. Since mid-2026 pursuing the degree full-time, including an Edge AI thesis on TensorRT inference optimization.

2017 — 2020

Apprenticeship · Physics Laboratory Technician

Primes GmbH · Pfungstadt

Vocational training in precision measurement, laser systems, optics, and technical documentation.

Skills

Machine Learning & AI

Python PyTorch ONNX TensorRT Edge AI LLM Integration AutoML

Data & Analytics

Microsoft PowerBI JMP (Statistics) Pandas / NumPy SQL Data Visualization

Systems & Dev

C++ NVIDIA Jetson Linux / systemd FastAPI Git Docker

// 02 · Projects

What I've Built

LLM · Autonomous Agent

Basti — 24/7 Discord AI

Full-featured Discord bot running as a systemd service on Jetson Orin Nano. Combines a music player (local MP3s + YouTube), queue management, audio filters, and an LLM personality layer with per-user memory, conversation history, web search, and image generation.

PythonGrok LLM discord.pysystemd Jetson

Agent · File System

Hermes — System Agent

Hermes agent on the Jetson handling file management, system tasks, and development automation. Includes a deep_coding skill built on the Anthropic API with prompt caching for cost-efficient multi-turn coding sessions.

PythonHermes Anthropic APIPrompt Caching Grok LLM

// 03 · Career Path

Where I'm Headed

2017 – 2021

Apprenticeship → Lab Technician · Primes GmbH

Vocational training as a physics laboratory technician followed by a first role in the field. Expertise in laser measurement systems, optics, and precise technical documentation.

2021 – 2026

Researcher · Procter & Gamble

Five years at a global Fortune 500 company — statistical analysis, PowerBI dashboards, consumer studies, and standardized test method development. Strong foundation in data-driven decision making.

2023 – 2026

B.Sc. AI & Machine Learning · part-time

Pursuing a degree in AI and Machine Learning at Wilhelm Büchner Hochschule alongside industry work. Deliberately building the technical foundation for a transition into ML engineering.

2026 – Present

Full-time Studies + Thesis + Portfolio ← you are here

Left industry to focus fully on completing the degree, working towards an Edge AI thesis (TensorRT inference optimization on NVIDIA Jetson), and building a hardware-grounded ML portfolio that demonstrates real deployment capability.

Targeting ML Engineer and Data Scientist roles in the Frankfurt area, combining a data analytics background from industry with hands-on ML and Edge AI engineering skills.

Long-term

Physical AI · Edge AI Specialization

Building toward robotics, autonomous systems, and embedded AI: domains where models run on constrained hardware and latency is a hard requirement, not a suggestion.