I’m Jonathan von Rad.

keywords = {

  • Multilingual NLP;
  • Large Language Models;
  • AI Safety;
  • Reinforcement Learning;
  • Model Efficiency;
  • Interpretability;
  • University College London;
  • University of Tübingen;
  • Peking University;
  • University of Tunis El Manar;
  • Foundations First;
  • Deutsch;
  • 中文;
  • العربية;
  • Iron Man (hopefully);
  • Basketball;
  • YouTube;

}

Portrait of Jonathan von Rad

London, UK

scroll

print(MySummary)

I recently completed my M.Sc. in Machine Learning at UCL, working on multilingual NLP, reinforcement learning and large language models.

My goal is to democratize AI along the multilinguality and efficiency axes. My research spans how models process and think across different languages, how they can be made smaller and cheaper without losing what matters and how we keep them safe, aligned and interpretable along the way. My most recent paper got accepted to EMNLP 2026 Main.

I’ve studied and worked in Germany, the UK, China and Tunisia and I speak German, English, Mandarin, Arabic and French. These experiences shaped what I care about most about: building language technology that is accessible for everyone, in every language and under every computational budget. My commitment to equitable and multilingual NLP helped me secure the highly competitive DAAD scholarship for my Master's in London, as well as a PhD offer from the University of Cambridge.

Education

.class GetDegrees

  1. Sep 2025 – Now · London, UK

    University College London

    M.Sc. Machine Learning

    Projected grade: Distinction (75%)

    • DAAD scholarship for master’s study abroad (10% acceptance rate; £41,000 total funding)
    • Coursework: Unsupervised Learning (Gatsby), Statistical NLP, Reinforcement Learning (DeepMind)
    • Thesis: Disentangling Tokenizer Quality from Script Disparity in Multilingual LLM Pretraining
  2. Sep 2021 – Aug 2025 · Tübingen, Germany

    University of Tübingen

    B.Sc. Computer Science

    First Class Honours (1.4 / 5.0, where 1.0 is the best) · Top 5% of undergraduate CS students

    • Coursework: Linear Algebra, Algorithms & Data Structures, Statistics, Artificial Intelligence
    • Thesis: Analyzing Pruning, Quantization and Distillation for LLM Compression (grade 1.0, GPA 4.0 / 4.0)
  3. Sep 2024 – Jun 2025 · Beijing, China

    Peking University 北京大学

    Exchange Student

    Studied Mandarin for one year (B2, HSK 5)

    • Merit-based Baden-Württemberg scholarship for two semesters abroad (3000€)
    • Research Assistant in LLM Alignment at the PKU Alignment and Interaction Lab
    • Second Place in Mandarin Speech Contest
  4. Mar 2023 – Dec 2023 · Tunis, Tunisia

    University of Tunis El Manar

    Visiting Student

    Studied Arabic (B2)

    • Studied Modern Standard Arabic and Tunisian Dialect within 9 Months
    • CS studies online (Theoretical Computer Science 2 and Artificial Intelligence)
  5. 2016 – Dec 2021 · Esslingen, Germany

    Schelztor Gymnasium Esslingen

    Visiting Student

    Abitur: 1.3 (GPA 4.0/4.0)

    • Majors in Maths, Physics and Sports; Minors in Computer Science and French

Publications

return Output

2026

EMNLP 2026 · Main Conference

UniComp: A Unified Evaluation Framework for LLM Compression via Pruning, Quantization and Distillation

Jonathan von Rad, Yong Cao, Andreas Geiger

2024

Experience

.class AddExperience

  1. Jul 2026 – Now · London, UK

    Foundations First Non-profit

    AI Solutions Engineer (Multilingual)

    • Developing AI-assisted differentiated learning to tackle education inequality in primary schools in Kenya
    • Using quantized Tiny-Aya 4B for on-device agentic modelling and multilingual NLU in Swahili
  2. Sep 2022 – Aug 2024 · Tübingen, Germany

    University of Tübingen

    Teaching Assistant, “Practical Programming I and II”

    Real Virtual Humans & Database Systems Group

    • Taught functional programming in DrRacket and object-oriented programming in Java (~25 students)
  3. Nov 2023 – Mar 2024 · Tübingen, Germany

    Coreflow UG

    Web Developer

    • Developed modular web applications for robot configuration using SvelteKit and TypeScript

Beyond Research

while(True)

  1. 2026

    Artificial Intelligence Review Springer Nature

    Reviewer

  2. Gemeindeverbund God in Life

    Youth Leader at my local church

  3. English Teaching

    Taught English to children in Beijing

Skills & Languages

import *

Technical

  • Python
  • PyTorch
  • NumPy
  • Pandas
  • Java
  • TypeScript
  • Linux
  • Git
  • R
  • LangGraph
  • LangChain

Languages

  • GermanNative
  • EnglishBilingual
  • MandarinAdvanced · HSK 5
  • ArabicAdvanced
  • FrenchIntermediate

YouTube

watch(JonnyInChina)

@JonnyInChina Videos from my YouTube channel.