Kajetan Knopp

Team Polar

Technologies: Autonomous Systems, AI, Environmental Modeling

Code availability: Proprietary

Description: Team Polar developed an autonomous navigation system for a research vehicle operating in Antarctica’s challenging terrain. The work involved simulation, optimization, and safety systems design for extreme environments.

My Role: Led the team and set the research direction. Managed the Autonomous Control subgroup and created the core logic and structure of the entire system. Engaged in environmental modeling, simulation, and path planning strategies. Collaborated with companies such as Nvidia and NXP, and presented our work at major events.

TKH AI Technology

Technologies: Python, Computer Vision, PyTorch, OpenCV, Industrial AI

Code availability: Proprietary

Description: TKH AI Technology is an AI consultancy and development firm within the TKH Group, applying machine learning and computer vision to real-world industrial challenges. The work spans client engagements across manufacturing, infrastructure, and smart systems.

My Role: Working as an AI Developer & Consultant, applying AI models to business problems with a specialization in Computer Vision. Responsible for model development, client communication, and translating technical solutions into production-ready systems.

Lie-Structured Shadow Tomography for Variational Quantum Circuits

Type: MSc Thesis - Quantum Computer Science, University of Amsterdam (2026)

Technologies: Python, Qiskit, Dynamical Lie Algebras, Variational Quantum Algorithms

Description: Constructs and analyzes a DLA-structured classical shadow protocol whose measurement unitaries and reconstruction map are adapted to the dynamical Lie algebra of a given parameterized quantum circuit (PQC) ansatz. Proves and numerically verifies that for specific relevant ansätze, this yields strictly better sample complexity for estimating VQA and QML observables than standard Pauli- or Clifford-based classical shadows.

My Role: Full research authorship - theoretical construction, proof development, and numerical experiments.

Differentiable Optimization for Cancer Modeling

Technologies: Julia, Cellular Potts Model (CPM), Stochastic Derivatives

Code availability: Open-source (Julia package)

Description: This research addresses the challenge of parameter optimization in stochastic lattice models, specifically the Cellular Potts Model (CPM), a widely used tool in cancer biology. CPM simulates the behavior of cells within tissues using probabilistic rules on a grid-based lattice.

My Role: Developed a novel gradient-based framework using automatic differentiation for discrete, stochastic models. The method overcomes the long-standing challenge of CPM’s non-differentiability. I applied it to prevent fingering instability, a phenomenon linked to cancer metastasis, by optimizing the model to suppress the formation of invasive cell structures.

This work lays the foundation for scalable, generalizable CPM simulations and enables the use of optimization techniques in previously inaccessible biological modeling domains.

Read Full Thesis

quantumz.io — Aircraft Allocation

Technologies: Python, Quantum Computing, Optimization, Operations Research, Aviation Systems

Code availability: Proprietary

Description: quantumz.io is a quantum-enhanced optimization startup. The project involved developing algorithms to optimize aircraft-to-route allocation for a large European airline carrier, minimizing costs and improving resilience to disruptions.

My Role: Quantum Intern responsible for designing and implementing optimization algorithms for the aircraft allocation problem. Work combined classical operations research with quantum-inspired approaches to improve scheduling efficiency under real-world operational constraints.

[Re] Interpreting CLIP with Hierarchical Sparse Autoencoders

Type: ML Research - Mechanistic Interpretability

Technologies: Python, PyTorch, CLIP, Sparse Autoencoders (SAE), Matryoshka SAE

Code availability: Open-source

Description: Reproduces and validates the core claims of Matryoshka Sparse Autoencoders (MSAE) for disentangling polysemantic CLIP representations. Proposes two novel extensions: a model-agnostic centroid-based method for assessing SAE neuron monosemanticity, and an in-depth evaluation of progressive recovery capabilities across varying levels of representational granularity.

My Role: Led reproduction study, designed and implemented both novel evaluation extensions, and co-authored the submission.

Deepsat

Website: deepsat.github.io

Technologies: Python

Code availability: Open-source

GitHub: github.com/deepsat

Description: Deepsat is a team project focused on developing an end-to-end pipeline for automatic terrain classification from mini-satellite imagery. The solution is being developed for a competition organized by the European Space Agency (ESA).

My Role: I implemented multiple ML models, including a ResNet50-based UNET for general semantic segmentation, an additional model using supplementary height data, and a YOLO-based pipeline for detecting smaller terrain features.

AI-Enabled Cardiology Care System

Program: EIT Health Digital Medical Devices Summer School - IESE Business School & University of Luxembourg

Technologies: Python, Machine Learning, Healthcare AI

Description: Developed an AI-enabled cardiology care system as part of a competitive cohort at the EIT Health Digital Medical Devices Summer School. The project combined clinical workflow understanding with ML-driven diagnostic support, evaluated against real healthcare delivery constraints at IESE Business School and the University of Luxembourg.

My Role: AI system design and implementation within a multidisciplinary team of healthcare and engineering professionals. Awarded EIT Health Alumnus status for contributions to digital healthcare innovation.

Picnic Technologies — Simulation Optimization

Technologies: Python, Simulation, Operations Research, Optimization, Stakeholder Management

Code availability: Proprietary

Description: Picnic Technologies is a European online grocery delivery company operating fully automated fulfillment centers. The Warehouse Simulation team models and optimizes the behavior of the entire warehouse system to improve throughput and reduce operational costs.

My Role: Worked as a 0.7 FTE analyst and developer in the Warehouse Simulation team. Responsible for optimizing productivity metrics, aligning with business stakeholders, and bridging the gap between engineering models and operational decision-making.

LLM Sustainability Tool - Treehacks (Stanford)

Event: Treehacks at Stanford University - attended with full travel scholarship

Team: Caltech, UC Berkeley, Stony Brook University

Technologies: Python, Large Language Models, NLP

Description: Developed an LLM-based sustainability analysis tool during Stanford's Treehacks hackathon. The system helps users and organizations understand the environmental footprint of decisions and products using language model reasoning over structured sustainability data.

My Role: LLM integration, prompt engineering, and system architecture.

Virtual Distillation - Mitiq

Technologies: Mitiq, Quantum Error Mitigation

Code availability: Open-source

Github: github.com/unitaryfund/mitiq

Description: Mitiq is an open-source Python library developed by the Unitary Fund for applying error mitigation techniques to quantum programs. It enables running noisy quantum circuits with improved output fidelity across a variety of hardware backends.

My Role: Implemented the Virtual Distillation method, a cutting-edge error mitigation technique that reduces noise in expectation values by simulating multiple copies of a quantum state. This approach is particularly effective for mitigating errors on NISQ devices.

BSO Toolbox

Technologies: Optimization, Natural Lighting, Computational Modeling

Code availability: Proprietary

Description: The BSO Toolbox is a research-driven computational design platform developed as part of a joint NWO-funded project involving TU/e and Leiden University. It focuses on optimizing building design, especially for improving natural light distribution.

My Role: Served as a part-time student research assistant under the supervision of an associate professor. Developed a new module within the BSO Toolbox for optimizing natural lighting in building layouts. Contributed to two MSc projects, acting as the lead for their technological implementation.

Ischemic Stroke Patient Coordination - Philips

Technologies: Python, Machine Learning

Code availability: Proprietary

Description: Participated in the Philips Brabanthon, where the challenge involved developing a strategy for coordinating ischemic stroke patients on a national level.

My Role: Developed and optimized the core of the strategy, winning the competition and presenting our work at the Philips Brabanthon.

Schiphol – Autonomous Baggage Handling

Technologies: AI, Automation, Strategic Consulting

Code availability: Proprietary

Description: A forward-looking consultancy project developed during the Netherlands-Asia Honors Summer School (NAHSS), focused on transforming baggage handling at Schiphol Airport using automation and AI. The initiative aimed to reduce delays, optimize routing, and improve passenger experience through intelligent systems.

My Role: Contributed to the strategic planning and system design. Proposed AI-driven solutions to automate baggage routing and reduce handling errors. Worked collaboratively on logistical modeling and human-technology integration within large-scale transport hubs.

ESGApp

Technologies: Django, Python, HTML, CSS, JavaScript

Code availability: Proprietary

Description: ESGApp is an intelligent recommendation platform that helps users discover mobile applications promoting sustainable and socially responsible behaviors. Developed for Bank Gospodarstwa Krajowego (BGK), the app includes ML-driven filtering and app verification mechanisms to ensure trust and transparency.

My Role: I developed all of the machine learning models, including:

• Comment classification for troll detection
• Category prediction for app relevance
• Malware scanning of Android apps

I also implemented the company verification module, parts of the backend using Django, and contributed to database schema design and infrastructure.

BraveBuddy

Event: TU/e Contest - Selected finalist, startup development program

Technologies: Python, Computer Vision, Eye-Tracking, Autonomous game logic

Code availability: Proprietary

Description: An autonomous eye-tracking game designed to ease children's anxiety during medical procedures. The system tracks gaze in real time to drive gameplay, keeping young patients engaged and calm without requiring any physical interaction. Developed during the TU/e Contest with a cross-disciplinary team, followed by participation in a startup development program.

My Role: Led computer vision and eye-tracking system development; designed core game interaction logic.

Student Focus Enhancement System - Hack the North

Event: Hack the North at University of Waterloo - attended with travel scholarship

Technologies: Python, LLM, Computer Vision, Vision-Language Models

Description: Built an LLM-vision hybrid system that monitors student engagement and focus during study or class sessions. Combines real-time computer vision for attention state detection with language model reasoning to generate personalized, context-aware interventions to re-engage distracted learners.

My Role: Computer vision pipeline, vision-language model integration, and real-time inference optimization.

LOT Employee App

Technologies: PowerApps, Azure

Code availability: Proprietary

Description: An enterprise app designed for office employees at LOT Polish Airlines, providing a suite of tools and modules for daily productivity, notifications, and services integration.

My Role: I developed one of the functional modules and built the associated Azure ML models used within the app's infrastructure.

Summario

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Proprietary

Description: Summario is a lightweight yet powerful solution for automated transcription and summarization of online meetings. It provides users with structured outputs from audio or video calls, allowing for better documentation and productivity.

What I Did: I developed the backend in Flask, designed and managed the database schema, and implemented core ML models for both transcription and summarization.

Link: Hackyeah Website

Quantum Intern - QuTech

Technologies: Quantum Computing, Reinforcement Learning

Code availability: Proprietary

Description: Internship at QuTech, the quantum research center of TU Delft and TNO. Work focused on the intersection of quantum computing and machine learning.

My Role: Developed a reinforcement learning framework aimed at optimizing quantum circuits, under the direct guidance of QuTech researchers.

Freetherapy

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Proprietary

Description: Freetherapy is a secure online platform built to facilitate group therapy sessions for mental health professionals and patients. It was designed as part of an initiative presented at the European Health Forum Gastein, in collaboration with EIT Health. The platform emphasizes privacy, simplicity, and accessibility.

My Role: I was responsible for most of the backend development, including the Flask server logic and database management. I also contributed to the session control and authentication systems.

Kagome Lattice Optimization

Technologies: Python, qiskit

Code availability: Proprietary

Description: Implemented quantum circuit optimization on the Kagome lattice on quantum hardware.

My Role: I implemented code and worked on the methodology - particularly focusing on the optimization algorithms for the quantum hardware.

Intangible Technology Transfers in Quantum Computing: Navigating EU Dual-Use Export Controls

Type: BA Thesis - European Studies (EU Law), University of Amsterdam (2026)

Technologies: Doctrinal legal analysis, law-in-context methodology

Description: Investigates the legal and practical constraints imposed on quantum research by the expanded EU Dual-Use Regulation (Delegated Regulation 2025/2003). Analyzes new export control classifications and evaluates how strict regulations on Intangible Technology Transfer (ITT) impact cross-border collaboration, cloud-based quantum computing access, and broader quantum innovation within the European Union.

My Role: Full research authorship - legal analysis, doctrinal methodology, and policy recommendations.

Social Network Analysis in Educational Psychology

Status: Published article (2025)

Technologies: Python, R, Complex Networks, Graph Theory

Description: Presents the possibilities offered by Social Network Analysis (SNA) in educational psychology research. The first part discusses the general principles of the network approach with a focus on social networks, followed by a review of network data features and the most important analytical techniques - both descriptive and inferential - applicable in educational psychology contexts.

My Role: Co-authored the article, contributed analysis methodology and implementation.

Stria Labs, INC

Technologies: Firmware, Embedded Systems

Code availability: Proprietary

Description: Stria Labs is focused on developing assistive technology for visually impaired users. The devices are designed to help users avoid obstacles and improve navigation in real-world environments.

My Role: Authored foundational firmware for the obstacle avoidance devices, enabling core functionality and hardware integration.

Netherlands-Asia Honors Summer School (NAHSS)

Type: Academic & Cultural Exchange

Location: Hong Kong & Mainland China

Description: Selected among top Dutch university students to participate in the prestigious NAHSS program, aimed at strengthening academic, cultural, and economic ties between Europe and Asia.

My Role: Participated in cross-cultural seminars, collaborated on international research challenges, and represented the Netherlands in academic and policy dialogues with Asian institutions and companies.

Modeling the Microbiome with Complex Networks

Website: networkpages.nl

Technologies: Python, R

Code availability: Proprietary

Description: As part of the Honors Academy at TU Eindhoven, I performed research into complex networks and graph theory - particularly how we can apply these techniques to analyzing the microbiome.

My Role: I implemented code, presented findings, wrote research reports, etc.

Code4Ukraine

Website: nonprofit.org.pl

Description: One of the first and largest hackathons aiming to provide humanitarian aid to those affected by the war in Ukraine. Organized both online and physically in Eindhoven, the event brought together developers, activists, and innovators from around the world.

My Role: I organized, promoted, and led the entire event. I also acted as the host and delivered a live Flask development workshop.

SmarTest

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Open-source

GitHub: github.com/SmarTest-nuevahacks/smarTest

Description: A secure, anti-cheating platform for online exams. Features automated cheating detection powered by ML models and server-side monitoring.

My Role: Developed the full backend including all Flask code, ML models, and database integration.

Code4Change

Technologies: PHP, HTML, CSS, JavaScript

Code availability: Both proprietary and open-source

GitLab: gitlab.com/code4change

Description: Code4Change is a platform that connects school and university students with real-world nonprofit software projects. It enables students to develop production-grade applications for charitable organizations, building both skills and social impact.

My Role: I was responsible for most of the backend development (in PHP), handled frontend components, created multilingual translations, and contributed to graphic and UI assets.

Botique

Technologies: Python, HTML, CSS, PHP

Code availability: Proprietary

Description: Botique is a SaaS platform offering small and mid-sized businesses tailored automation solutions (bots) to eliminate repetitive manual tasks. The service emphasizes affordability, ease of use, and value-driven automation to boost operational efficiency.

My Role: I led the full development of Botique, implementing all backend automation logic, integrating third-party APIs, and building the frontend user interface and dashboard from scratch.

HealthExchange

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Open-source

GitHub: github.com/khknopp/healthexchange

Description: HealthExchange is an open-source platform created to support collaboration between hospitals and healthcare institutions during the COVID-19 pandemic. It allows facilities to offer, request, and exchange medical equipment based on need and availability.

My Role: I independently developed the full platform, including the backend in Flask, database logic, and frontend implementation. The project focused on providing a lightweight, efficient, and easy-to-use interface during an urgent global crisis.

SwiftPlan

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Open-source

GitHub: github.com/swiftplan-winhacks/swiftplan

Description: SwiftPlan is a web-based scheduling platform that helps groups optimize their meeting times by minimizing idle time between individual commitments. It uses a custom-built heuristic algorithm to propose optimal group slots, improving calendar efficiency for busy users.

My Role: I developed the frontend interface and database schema, and contributed to backend logic in Flask. My focus was on intuitive UX design and robust data integration for cross-calendar syncing.

Corominder

Technologies: Flutter

Code availability: Open-source

GitHub: github.com/khknopp/corominder

Description: Corominder is an open-source mobile application designed to promote public health during the COVID-19 pandemic. It delivers real-time news updates via RSS feeds and reminds users to maintain hygiene and follow safety protocols.

My Role: I implemented the RSS feed integration to provide live news updates on COVID-19 developments, leveraging Flutter's capabilities for efficient data rendering.

FF9M

Technologies: Python, PHP, HTML

Code availability: Proprietary

Description: FF9M is a comprehensive, custom-built software suite developed for a local film festival. It includes modules for participant registration, film submission, audience voting, and data analytics to support real-time decision-making during the event.

My Role: I independently designed and implemented the entire platform, including front-end forms, back-end logic in Python and PHP, and analytical dashboards for festival administrators.

PruGym

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Open-source

GitHub: github.com/Hadamard-WawCode/PruGym

Description: PruGym is a lightweight web application that allows users to quickly find the nearest sports facilities in Warsaw. It integrates real-time location filtering with a clean and responsive frontend to help residents and visitors locate gyms, courts, and fitness centers.

My Role: I contributed to the design and implementation of the database schema and backend functionality using Flask. I focused on data integration and API logic to ensure fast and relevant facility lookups.

PrudensBin

Technologies: Flask, Python, HTML, CSS, JavaScript

Code availability: Open-source

GitHub: github.com/PrudensBin-Hackyeah/PrudensBin

Description: PrudensBin is an eco-focused web application that helps users locate the nearest disposal facilities for standard and non-standard waste types. From couches to batteries to expired medicine, the platform educates and guides responsible recycling in local communities.

My Role: I was responsible for a large part of the backend logic and Flask routes, contributed to the frontend interface, and co-developed the database structure and data mapping logic.

Bayer Hackathon

Technologies: Python, Flask

Code availability: Open-source

GitHub: github.com/bayerhealth/bayerhealth

Description: An intelligent system designed during the Bayer Hackathon to assess the health and morphology of various crops using computer vision and predictive analytics. The platform provides actionable insights for agricultural optimization and crop management.

My Role: I designed and implemented the machine learning models responsible for plant health evaluation, and contributed to the backend architecture using Flask.