I'M
Toby
Kreiman.
Computer Science and Physics at Columbia University
*this website is out of date, please see updated site here*
01 PORTFOLIO
MY LATEST WORK. SEE MORE >
02 Experience
Sep 2021 - Present
Ciocarlie ROAM Lab Columbia University
Reinforcement Learning Researcher
-
Developed deep reinforcement neural networks to solve problems of robotic grasping
and in hand manipulation
-
Worked on novel Hardware as Policy techniques to co-optimize hardware and
computational policy
Jun - Aug 2022
Machine Learning Intern
-
Embedded videos using multimodal joint audio-text model to learn musical genre
-
Developed an online learning binary classifier to improve music autoplay
recommendations on YouTube based on embeddings
Jun - Aug 2021
Meta (Facebook)
Machine Learning Intern
-
Trained novel machine learning models based on these features and billions of data points to recommend new podcast content
-
Engineered features from visual, audio, and natural language text data
-
Created new feature extraction pipeline that improved efficiency by 10x based on
previous approaches
Jun - Aug 2020
Supportiv
Machine Learning Intern
-
Created both supervised and unsupervised machine learning models to classify and summarize text
-
Analyzed modern literature dealing with transformer models and used transfer learning to improve accuracy on conversational dataset
-
Constructed a Name Entity Recognition (NER) model to accelerate knowledge extraction pipeline
Summer 2017, 2018
Poggio Lab Center For Brains Minds and Machines MIT
Machine Learning Research Intrern
-
Developed new reinforcement learning algorithms to tackle “catastrophic forgetting” in
neural networks, resulting in publication in peer reviewed journal
-
Created efficient convolutional based neural networks to tackle the problem of object
detection and feature point extraction on a mobile device
-
Constructed a mobile application that created a 3D map around the user and
communicated via vibrations, allowing a blind person to navigate
