Motivational Letter
Dear Hiring Manager and the Applied AI Team at UMCG,
Welcome to my application. I built this interactive website over a weekend for one reason: to demonstrate my passion for building practical, full-stack AI solutions and my ability to rapidly learn and deploy new technologies.
In my current role, my focus is always on creating tangible value by reducing cognitive load and streamlining complex workflows. I've developed an automated transcription service that turns support calls into structured tickets and built an intelligent support bot using LangChain and RAG to give my colleagues instant access to knowledge. While I am proud of these technical achievements, I am most energized when I can innovate and experiment. This drive to build and take initiative is why the description of your team attracted me (as a place where "ideas and initiative are truly valued"). My recent conversation with Akshay Biere confirmed this, and I am incredibly excited by the prospect of joining an optimistic and forward-thinking team that is not afraid to try and fail together.
My desire to join UMCG is also a deliberate choice to align my technical skills with work that is unambiguously positive. My career has always balanced a passion for technology with a focus on people, from my studies in Psychology to my work in a client-facing support role. I thrive on helping others, and I am looking for an environment where that help has a clear and meaningful impact. The opportunity to develop solutions that directly support patients, assist clinicians, or accelerate medical research offers a sense of purpose that is uniquely compelling.
This website itself is a testament to my capabilities. To build it, I taught myself Svelte (a framework mentioned in your vacancy that I had no experience with) and integrated a Flask backend with an OpenAI-powered chat via WebSockets. It demonstrates my full-stack proficiency and my commitment to continuous learning. While I haven't worked extensively with Microsoft Azure yet, my quick grasp of Svelte and my experience with Docker and cloud deployments make me confident that I can master the Azure AI stack swiftly. I am eager to dive into Azure Machine Learning and OpenAI services and am prepared to pursue certifications to formalize that knowledge.
I am confident that my practical skills, learning agility, and genuine drive to contribute to smart healthcare will enable me to be a valuable member of your team from day one.
Thank you for your time and consideration.
Met vriendelijke groet (With kind regards), Sander Kaatee
P.S. I invite you to interact with the AI chat on this site. I built it as a practical way to demonstrate the skills mentioned in your vacancy, from prompt engineering to API integration. Jokingly, it's ready to conduct the first round of the interview. It's a small example of the hands-on, experimental approach I'm excited to bring to the Applied AI team.
Ask Me Anything
Powered by GPT-4.1-mini
This AI may occasionally provide inaccurate information. Please verify important details with Sander directly for accuracy.
Resume
Download PDFLoading resume...
Technical Portfolio
Enterprise Solutions
At my current role, I focus on creating practical AI solutions that reduce cognitive load and streamline workflows. While I cannot share the code publicly, I'm happy to demonstrate these projects live.
Automated Transcription Service
Developed an intelligent system that automatically transcribes support calls and converts them into structured tickets, significantly reducing manual workload for the support team.
RAG-Powered Knowledge Base
Built an intelligent support bot using Retrieval-Augmented Generation to give colleagues instant access to company knowledge, improving response times and consistency.
Competition Achievements
Engineered a class-aware segmentation ensemble for 12-band Sentinel-2 imagery. Trained a custom MSNet (ResNet-50 encoder + pixel-shuffle decoder) in PyTorch Lightning with mixed-precision and Dice + Soft-BCE loss. Used Bayesian optimization via scikit-optimize for augmentation, sampling, and LR schedule, guided by permutation importance. Built class specialists for mining, logging, and shrubland, merging polygon predictions into a single GeoJSON via a lightweight ensembler—boosting F1 and landing in the top 8%.
View results →Re-engineered the starter pipeline into a lightweight 1-D ResNet that ingests 150-band spectra and directly regresses P, K, Mg and pH. Three independently seeded models were averaged to tame variance; Bayesian hyper-parameter search (scikit-optimize) and aggressive early-stopping shaved ≈20 % off the public-baseline RMSE without extra metadata or handcrafted features.
View results →Open Source Projects
Interactive Recruiter Site (This Website!)
🚀 You're viewing this project right now! Built specifically for UMCG application.
Built this full-stack application over a weekend to demonstrate my current skills. Features real-time AI chat, multilingual support, and modern responsive design.
Key Achievement: Learned SvelteKit from scratch over a weekend and integrated it with a Flask backend, demonstrating rapid technology adoption and full-stack capabilities.
Trickie.me
A playful twist on Dutch payment requests with gamification. Full-stack Flask app with Canvas-based wheel game, Docker deployment, and responsive design.
SomethingLikeMeGen
Automatic music video generator using computer vision. Analyzes movies for faces, creates scene-aware cuts, and syncs to audio tracks.
GPT Plays Pokemon
Built an autonomous Pokémon Red player using GPT-4 Vision right after launch (back in 2023). AI analyzes screenshots, sets hierarchical goals, and generates button sequences to play end-to-end. Originally considered live streaming on Twitch/YouTube but hesitated due to cost concerns and doubts about GPT-4's gaming abilities. Now with 'Claude Plays Pokémon' trending on Twitch, it's clear this was a missed opportunity. Still, it's a great real-world highlight of applied AI.
Explore more projects in my GitHub portfolio →