LinkedIn: jacob-ungar-felding
GitHub: jfelding
Email: jfelding@gmail.com
- MSc in Computational Physics, Niels Bohr Institute, Copenhagen (2019–2021)
GPA: 11.7/12.0
High-performance computing, C++, large-scale data analysis, ML fundamentals, fluid dynamics, scientific computing, optimization, image processing. - BSc in Physics, Niels Bohr Institute, Copenhagen (2015–2018)
GPA: 10.6/12.0
Theoretical & experimental physics, applied statistics & ML coursework.
BA project: Detection using hyperspectral imaging & CNNs
- The Scalable Spatial Echo State Network for Detection of Anomalous Ocean Surface Topography
Supervisor: James Avery (avery@nbi.dk)
Description: Developed a linearly-scalable echo-state RNN for high-dimensional spatio-temporal forecasting without knowing the governing PDEs. Built an adaptive anomaly detector for ocean transitions, training on years of simulation data in minutes with one-shot (no gradient) updates.
Grade: 12 / A
Read the thesis
- Metric learning, embedding & classification models
- Object detection; synthetic data (diffusion) pipelines
- Vector search & anti-spoofing; continuous model training, deployment & validation
- End-to-end production integration; internal tooling & customer-facing web apps (SPA)
- Computer vision for fraud detection & document processing
- Offline signature verification; handwriting segmentation & extraction
- ML Deployment & MLOps; SQL data engineering; full-stack software development (Python, JS)
- Designed & optimized image-processing algorithms for multispectral imaging (e.g. pixel-level classification)
- Parallelized workflows; interfaced with hardware triggers
- UI development & DevOps improvements; customer support & training
- Technologies: C#, .NET, Python
- Managed transactions & customer inquiries
- Tutored first-year physics students in Python & computational methods
- Administrative support: scheduling, document prep, student liaison
- A New ML Method for High-Dimensional Systems
Enhanced Echo State Networks for million-variable spatio-temporal data (remote sensing, anomaly detection). Achieved near-linear training time with one-shot updates. - A Tunable Loss Function for Image / Video Data
Extended IMED (Image Euclidean Distance) to FFT/DCT versions for RGB, hyperspectral & MRI; added robust regression support. - Detecting Frost-Damaged Sugar Beets
Hyperspectral line-scan pipeline + CNN classification in collaboration with Nordzucker/NEWTEC. - Oral Cavity Cancer Recognition
Preprocessing (morphology, Fourier), windowed labeling & ML pipeline for dental imaging (KU/SDU). - Time Series Analysis for Airlabs
(S)ARIMA forecasting of air-quality measurements in Copenhagen; sensor calibration & real-world deployment.
- Languages: Python, C#, C++, TypeScript, JavaScript, MATLAB
- Frameworks & Tools: TensorFlow, PyTorch, OpenCV, Docker, Azure DevOps, Jamstack, Cloudflare Workers, MinIO
- Data & MLOps: SQL, pandas, NumPy, CI/CD, model deployment
- Version Control: Git, SVN
- Platforms: Linux, Mac, Windows (if I must)
- Danish: Native
- English: Academic (TOEFL 112/120)
- German: Basic
- Computer Vision & Algorithm Engineering
- Machine Learning & Data Analysis
- High-Performance Computing & Production Software Development
- Emerging Technologies
- Hydroponics: Soil-less gardening & controlled cultivation
- Music: Choir singing & piano
- Cooking & Baking: Gadget-driven recipe optimization
- Repairs & Maker Culture: Bike & electronics repair; 3D-printed replacement parts; Repair Café volunteer
- Den Georgbruunske Pris (2022): DKK 100,000 award for uncovering overbilling at Charlottenlund Lægehus
- National History Competition Winner (2012): And Körber-Stiftuing EUSTORY allum