Georg Lange 🧠
Georg Lange Georg Lange

Independent Researcher

Biography

I’m an independent researcher working on Mechanistic Interpretability for LLMs. I was a MATS scholar and worked with Alex Makelov and Neel Nanda on Sparse Autoencoders and Distributed Alignment Search for feature detection and subspace activation patching. Previously, I was a MSc AI student at the University of Amsterdam, where I worked on brain-like interpretable spatiotemporal Computer Vision models, supervised by Prof Iris Groen and Amber Brands.

Further, I was a graduate student at the Graduate Center, CUNY, where I worked on Reinforcement Learning, Decision Making, and Reward Sensitization and conducted fiber photometry experiments in the Nucleus Accumbens of mice, supervised by Prof Jeff Beeler.

Download CV
Interests
  • Artificial Intelligence
  • Mechanistic Interpretability
  • Systems Neuroscience
Education
  • M.Sc. Artificial Intelligence

    University of Amsterdam

  • M.Sc. Cognitive Neuroscience

    Graduate Center, City University of New York

  • B.Sc. IT-Systems Engineering

    Hasso-Plattner-Institut, Potsdam

Featured Publications
Recent Publications
Projects

Experience

  1. Scholar (ML Alignment Theory)

    SERIMATS (Stanford Existential Risk Initiative)
    • Researched Mechanistic Interpretability for LLMs with Alex Makelov, mentored by Neel Nanda (3 months full-time, 4 months part-time)
    • Worked on Sparse Autoencoders and Distributed Alignment Search for feature detection and subspace activation patching
  2. Consultant for Data Science and Cloud Computing

    Datametric
    • Developed Data Science solutions for major companies (Vattenfall, Ikea, T-Mobile, Vesting-finance) using AWS, Sagemaker, and Azure ML (part-time)
  3. Student Consultant

    SUGAR Network, HPI, KIT
    • Developed a data-driven product from scratch for a major German insurance company
    • Applied Design Thinking and data-driven decision-making
    • Developed an application to support the product
  4. AI Engineer (Intern and Working Student)

    QiO Technologies Ltd.
    • Led an AI project for a British water company
    • Built an end-to-end computer vision pipeline for sewer damage detection using transfer learning, data cleaning, hyperparameter tuning, and image visualization
  5. Research Assistant

    Digital Health Center, Hasso-Plattner-Institute
    • Developed the frontend for a molecular tumor board
    • Designed a psychological test battery for epilepsy research
  6. Research Assistant

    Digital Health Center, Hasso-Plattner-Institute
    • Continued work on the frontend for a molecular tumor board
    • Enhanced the psychological test battery for epilepsy research

Education

  1. M.Sc. Artificial Intelligence

    University of Amsterdam
    • Visited in-depth courses about Machine Learning, Deep Learning, Information Retrieval and did a research project about Privacy in complex-valued DNNs
    • Research in complex-valued neural networks for privacy protection and equivariant spatiotemporal CNNs for scene representation learning
    • Thesis on brain-like interpretable spatiotemporal Computer Vision models with adaptation mechanisms, supervised by Prof Iris Groen and Amber Brands
  2. M.Sc. Cognitive Neuroscience

    Graduate Center, City University of New York
    • Investigating neural correlates of motivation and effort-based decision making in hyperdopaminergic (DAT-KD) and conditional D2-receptor KO (fDRD2 x Adora2a::Cre) mice using fiber photometry in Nucleus Accumbens
    • Thesis on interaction between dopamine and acetylcholine during cocaine- or amphetamine-induced drug sensitization, supervised by Prof Jeff Beeler
    • Developed Fibermagic, a software package for fiber photometry data analysis
    • Developed Magicbox, a platform to control operand boxes and neural data acquisition systems wirelessly
    • Hired and trained two undergraduate students on experimental neurobiology
  3. B.Sc. IT-Systems Engineering

    Hasso-Plattner-Institut, Potsdam
    • Created an online course on Deep Learning for Computer Vision for 13k participants
    • Developed an Android App, Backend, and Dashboard for unobstrusive health monitoring of wearable devices for connected healthcare
    • Thesis on “Detecting Several Types of Distractions During Work Using Wireless EEG-Devices Applying Machine Learning Techniques”, supervised by Prof. Dr. Bert Arnrich
Skills
Technical Skills
Python

Pandas, NumPy, SciPy, Sklearn, Matplotlib, Plotly, Seaborn, Statsmodels

PyTorch

Lightning, TensorFlow, WandB

Cloud Computing

AWS, Sagemaker, Athena, EMR

Software Development

Kotlin, Java, SQL, Arduino, 3D Printing, Laser Cutting, Design Thinking

AI Research Skills
Mechanistic Interpretability

Sparse Autoencoders, transformer-lens, LLMs

Deep Learning

WandB, PyTorch, Einops, Einsum

Data Analysis & Visualization

Pandas, Matplotlib, Plotly, Seaborn

Neuroscience Skills
Experimental Techniques

Stereotactic brain surgery, viral injections, fiber implantation, dual-color fiber photometry

Animal Handling & Behavioral Studies

Transgenic mouse colony management, PCR genotyping, conditioning, operand boxes, IP-injections

Histology & Neural Analysis

Perfusion, cryostatic/vibratome slicing, immunohistochemistry, EEG data analysis

Custom Behavioral Setup Development

3D-printing, computer vision, software development, electronics, raspberry pi

Awards
Research Grants for SERIMATS
AI Safety Support ∙ June 2023
Received research grants for participation in SERIMATS and its extension, supporting AI alignment research.
Fulbright Scholar
German-American Fulbright Program ∙ February 2021
Awarded a scholarship to support studies and research in the United States, fostering transatlantic academic exchange.
Research Award & Assistantship
Cognitive Neuroscience Program, Graduate Center, CUNY ∙ April 2022
Recognized for research contributions with awards and assistantship in the Cognitive Neuroscience program.
Scholarship from Konrad-Adenauer-Foundation
Konrad-Adenauer-Foundation ∙ January 2018
Financial stipend covering living costs; participated in seminars and events focused on leadership and policy.
Online Course Deep Learning for Computer Vision
openHPI ∙ October 2019
Created and led a four-week MOOC on Deep Learning and Image Recognition with over 13,000 enrollments.
Hack Zurich Challenge Winner
Hack Zurich ∙ April 2020
Winner of Europe’s largest hackathon, Hack Zurich, competing with over 1,000 participants.
Organizer of Senkrechtstarter Program
Senkrechtstarter ∙ February 2020
Founded and organized a mentoring program connecting students with pupils from non-academic backgrounds.