Filip Radenović
Research Scientist at Meta
Montenegro
Summary
🤩
Rockstar🎓
Top SchoolFilip Radenović is a Research Scientist at Meta Superintelligence Labs with eight years of experience at the intersection of LLMs, perception, and multimodal media understanding and generation. He holds a PhD in Computer Vision and spent six years in academic research at the Center for Machine Perception before joining Meta, combining deep theoretical grounding with applied industrial R&D. His work spans large language models and visual feature learning, and he has contributed notable open-source tooling for CNN-based image retrieval in PyTorch, emphasizing practical model optimization and novel pooling/loss layers. Based in Montenegro, he brings an international perspective to US-based labs and has a track record of moving research code toward robust, production-ready implementations. Colleagues describe him as someone who blends rigorous signal-processing roots with modern deep learning practice, often surfacing subtle numerical fixes and tooling improvements that improve reproducibility.
8 years of coding experience
6 years of employment as a software developer
PhD Computer Vision, PhD Computer Vision at Czech Technical University in PragueGrammar School "Slobodan Škerović", PodgoricaMSc Electronics Telecommunications and Computer Sciences, MSc Electronics Telecommunications and Computer Sciences at Faculty of Electrical Engineering, University of Montenegro
serbian/montenegrin, English
Github Skills (13)
mask-rcnn10
faster-rcnn10
pytorch10
machine-learning10
convolutional-neural-networks10
python10
image-retrieval10
data-structure9
algorithm9
data-structures9
algorithms9
computer-vision8
linear-algebra8
Programming languages (6)
Github contributions (5)
filipradenovic/cnnimageretrieval-pytorch
Apr 2018 - Jun 2022
CNN Image Retrieval in PyTorch: Training and evaluating CNNs for Image Retrieval in PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:3 releases, 1 review, 62 commits in 4 years 1 month
Contributions summary:Filip primarily contributed to the development and improvement of the CNN image retrieval system. Their work involved fixing string formatting, adding pretrained networks with automatic download functionality, and fixing potential issues with Cholesky decomposition in the whitening process. They also added new layers, loss functions, and various pooling options, including a regional pooling layer, highlighting a focus on feature extraction and model optimization for image retrieval tasks within the PyTorch framework.
pytorchpythondeep-learningcnnstraining
filipradenovic/revisitop
Apr 2018 - Jun 2022
Revisiting Oxford and Paris: Large-Scale Image Retrieval Benchmarking
Contributions:40 commits, 2 PRs, 31 pushes in 4 years 2 months
pythonbenchmarkingmatlabmriscale
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.