Daniel Dager

Research internship under the supervision of Yair Lakretz and Emmanuel Chemla, Laboratoire de Sciences Cognitives et Psycholinguistique, to investigate the validity of the dual-stream model of speech processing and production. We are building a multimodal autoencoder to simulate speech production in auditory and visual tasks, later to be compared to behavior data in human subjects to be collected in a series of experiments we will conduct in early 2025. Research internship under the supervision of Daniel Pressnitzer, Laboratoire des Systèmes Perceptifs, Department d'Études Cognitives, investigating the ability of humans subjects to exhibit random behavior. Designed and carried out an experiment where subjects were asked to generate random sequences of numbers while collecting various physiological data. Trained various neural network models on said data to assess both the randomness of the sequences produced, and the ability of the models to generalize to new subjects.

Automated the aggregation of a database of faculty and administrators in all 50 states. Made significant frontend and backend contributions to the launching of Coursicle 2.0, most notably to Courisicle Chat, a messaging platform that services tens of thousands of students.

Assisted in the deployment of a local speech-to-text model for near perfect audio transcription. Designed and implemented the company’s first AI-focused content generation tool suite. Launched audio denoising service that used machine learning to enhance speech-audio quality.

Optimized the lab’s flagship algorithm, MAGIC, which restores large datasets of single-cell RNA sequences via interpolation by diffusion. Studied the relationship between linear algebra and stochastic processes, how to implement linear operations efficiently, and how to visualize multi-dimensional datasets.

I worked at the John P. Hussman Institute for Human Genomics under Dr. Derek Dykxhoorn, PhD. Studied the processes by which researchers generate induced pluripotent stem cells by subjecting target cells to certain reprogramming factors.