Martin Weiss Email and Phone Number

- Leveraged YOLOv7 to detect wireless signals from spectrogram images- Developed custom YOLOv7 based models to maintain spatial alignment of input images- Extracted features from detected signals for further analysis- Applied UMAP for dimensionality reduction and outlier detection from extracted features

- Fine-tuned Google’s FLAN-T5 LLM to create dictionaries from sequencing workflow comments- Developed models using PyTorch to predict 2D histograms using sequencing run metadata- Achieved >99% accuracy in identifying the 0.25% of sequencing runs that were aberrant