About Me

• I am a Master Student in Artificial Intelligence at University of Bern, under the umbrella of CAIM and ARTORG.
• I am currently working on clustering algorithms for subtype and stage inference for FDG-PET images of Amyotrophic Lateral Sclerosis patients, under the supervision of Prof. Dr. Kuangyu Shi and Postdoc. Jimin Hong.
• I have received a Bachelor’s degree in Computer Engineering at Technical University of Cluj-Napoca, Romania. My dissertation was based on Self-Supervised Learning for Heart Arrythmia Detection and Classification, advised by prof. Cristian Vicas.
• During my bachelor, I interned at Siemens, mentored by Irina Emilia Nicolae, where I discovered my passion for Artificial Intelligence while working on Machine Learning methods for EKG data.

Projects

• Image Captioning

Encoder-Decoder architecture for generating captions

Technologies used: Dinov2, Transformers, PyTorch, Wandb.

View on GitHub

• LegalBert Text Summarization

Quantized INT8 method for LegalBert inference

Technologies used are PyTorch, Transformers,BitNet, BERT, HuggingFace, Wandb.

View on GitHub

• Medical Corpus NLP

Named-Entity Recognition for Custom Labels

Technologies used are Spacy, NLTK, BERT, GPT, Label-Studio, Wandb, HuggingFace.

View on GitHub

• Bladder Cancer Proteomics

Subpopulation and Feature Detection

Given a clinical dataset with bladder cancer patients, we identify 7 subpopulations. We infer from proteomics and metadata, the most informative features.

View on GitHub

• Pancreatic Cancer Genomics

K-Means Clustering

Pancreatic Cancer Risk Genes simulated dataset using R Studio and additional plots for GWAS study plots.

View on GitHub

• Computer-Assisted Surgery Pipeline

Preoperative Planning

We apply pivot calibration, Region-Growing segmentation with seed, Point Cloud Registration and U-NET segmentation deep learning for reference.

View on GitHub

• Mini NERF

Bulldozer Neural Radial Field

Based on NERF paper, we reconstruct the point cloud from views of complex scenes using 5D coordinates.

View on GitHub

• Mountain Car DQN

Deep Q-Learning and REINFORCE

Reinforcement Learning algorithms for model-free control of cart pole in openai gym.

View on GitHub