Nikolaj Banke Jensen

🇩🇰/🇬🇧 • Machine Learning, Programming Languages, Category Theory

I'm a student in computer science, primarily interested in machine learning, programming languages, and the intersection between the two. Machine learning is compositional in nature, so I'm exploring how type theory and category theory can help reason about these systems. Ultimately, I want to create better tools for working with, and proving properties of, compositional structures.

Want to get in contact? Email me at nikolaj@nikolajjensen.com.

Publications

Enforcing Idempotency in Neural Networks

with Jamie Vicary • ICML 2025

Paper

Poster

What I'm working on

Kepler

This is a C++ implementation of APL I wrote for my BSc dissertation. It is one of the most feature-rich open-source APL implementations available, supporting most primitive functions and operators. The following Kepler program sets up and runs 5 rounds of Conway's Game of Life!

⍝ Setting up a 10x10 board.
glider ← 3 3⍴1 1 1 1 0 0 0 1 0
board ← ¯10 ¯10↑glider

⍝ The function of life.
Life ← {1 ⍵∨.∧3 4=⊂+/,¯1 0 1∘.⊖¯1 0 1∘.⌽⊂⍵}

⍝ Helpers for updating the board.
Update ← {board ← Life board}
Round ← {Update ◊ ⎕←'\n \n' ◊ ⎕←board}

⍝ Run 5 rounds.
(Round⍣5)0

If you like functional programming, (semi-)esoteric languages, and funky symbols, you'll love Kepler.

See it on GitHub

SunCalc

A Swift Package for calculation of Sun and Moon positions and phases. Originally written in Java, I have rewritten this project as a native Swift Package for easy use on Apple platforms. This work is open source and is published under the Apache License 2.0.

See it on GitHub

Where I've been

DPhil Computer Science

University of Oxford • 2024 - present

Supervised by Sam Staton.

MPhil Advanced Computer Science

University of Cambridge • 2023 - 2024

Thesis titled "Enforcing idempotency in neural networks".

Supervised by Jamie Vicary.

BSc Computer Science (Artificial Intelligence)

King's College London • 2020 - 2023

Dissertation titled "Building Kepler: A Feature-First APL Implementation in C++".

Supervised by Laurence Tratt.

Positions I have held:

Research Assistant

University of Cambridge • Jul - Sep 2024

Developing gradient-free optimisation methods for finding neural networks which act as idempotent mappings. This work is a continuation of my MPhil thesis and is under supervision of Jamie Vicary.

Part-Time Undergraduate

Arm • Jun 2022 - Mar 2024

In the Compute Library team, working on fast ML for mobile and embedded Arm platforms. Focused on JIT compilation of Arm64 assembly for matrix multiplication.

Research Assistant

King's College London • Jun 2021 - Mar 2023

Worked on educational programming tools in the research group led by Michael Kölling. Helped design and conduct an eye-tracking study to assess the impact of background styling in code editors on novice programmers' code understanding. The latter resulted in a paper.