My name is Niels, I'm 28 years old and I live in Belgium.
I'm currently working as an ML engineer @ 🤗 HuggingFace, where I'm part of the Open-Source team.
I work on HuggingFace Transformers, a Python library implementing several state-of-the-art AI algorithms, all based on the original Transformer by Google.
I love making AI more accessible to anyone. So far, I've contributed the following algorithms to HuggingFace Transformers:
- TAPAS, by Google AI
- ViT, by Google AI
- DEiT, by Facebook AI
- DETR, by Facebook AI
- BEiT, by Microsoft Research
- CANINE, by Google AI
- LUKE, by Studio Ousia
- LayoutLMv2 and LayoutXLM, by Microsoft Research
- DINO, by Facebook AI
- TrOCR, by Microsoft Research
- SegFormer, by NVIDIA
- ImageGPT, by OpenAI
- Perceiver/Perceiver IO, by Deepmind
- MAE, by Facebook AI
- ViLT, by NAVER AI Lab
- ConvNeXT, by Facebook AI
- DiT, by Microsoft Research
- GLPN, by KAIST (Korea Advanced Institute of Science and Technology)
- DPT, by Intel Labs
- YOLOS, by School of EIC, Huazhong University of Science & Technology
- TAPEX by Microsoft Research
- LayoutLMv3, by Microsoft Research
- VideoMAE, by Multimedia Computing Group, Nanjing University
- Donut, by NAVER AI Lab
- X-CLIP, by Microsoft Research
- Deformable DETR, by SenseTime Research
- MarkupLM, by Microsoft Research
- LiLT, South China University of Technology
- Table Transformer, by Microsoft Research
- CLIPSeg, by University of Göttingen
- Audio Spectrogram Transformer, by MIT Computer Science and Artificial Intelligence Laboratory, Cambridge
- BiT, by Google AI
- ViT Hybrid, by Google AI
- Swin2SR, by CAIDAS, University of Würzburg
- GIT, by Microsoft Research
- UPerNet, by Peking University
- BLIP-2, by Salesforce
- InstructBLIP, by Salesforce
- FocalNet, by Microsoft Research
- PerSAM, by Shanghai Artificial Intelligence Laboratory
- DINOv2, by Meta AI
- ViTMatte, by HustVL
- Nougat, by Meta AI
- OWLv2, by Google AI
- SigLIP, by Google AI
- SlimSAM, by National University of Singapore
- Depth Anything, by The University of Hong Kong/TikTok
- UDOP, by Microsoft Research
- ZoeDepth, by Intel Research
Besides that, I help others add models to the library, including:
- Swin Transformer, by Microsoft Research
- mLUKE, by Studio Ousia
- Nyströmformer, by University of Wisconsin-Madison
- YOSO, by University of Wisconsin-Madison
- PoolFormer, by Sea AI Lab
- CvT, by Microsoft Research
- GroupViT, by NVIDIA
I'm mostly working with PyTorch. For IDEs, I work with Visual Studio Code and Google Colab. It's all you need.
I learned everything about deep learning through self-study, mainly thanks to Andrej Karpathy's cs231n course at Stanford University (lectures are free on Youtube and assigment solutions can be found on Github).