Daniel Levine Email & Phone Number

New York, Ny, Us

New Haven, Ct, Us

Postdoctoral Associate in the van Dijk Lab at Yale University, focusing on developing and applying deep learning methods to biological data.- Developed Cell2Sentence (C2S), the first single-cell foundation Large Language Model (LLM), published at ICML 2024- Contributed to the development of BrainLM, a foundation model for brain activity recordings- Developed CaLMFlow, a method bridging large language models and flow learning methods for modeling perturbations of single-cell data- Worked on innovative approaches combining numerical analysis and operator learning to improve neural network training

New York, Ny, Us

• Researched and developed new micro-services in Python to improve email security: decrypting email attachments, auto-discovery of on-premise MTAs, detection of phishing login pages, decrypting URLs, anomaly detection• Improved features and added new features to model, fine-tuned model, analyzed using A/B testing with F1/Recall/Precision metrics and area under ROC curve along with inference analysis for specific sub-populations• Performed research and experiments on state-of-the-art transformer models, gradient boosted models, NLP techniques, and topic modeling• Built micro-service to leverage asynchronous API calls to product for all customers• Assisted in client and sales calls by explaining model misses, handling bug fixes, and creating solutions to new email phishing attacks

Phoenix, Az, Us

• Worked through weekly projects building predictive models using python packages such as scikit-learn, statsmodels, and tensorflow/keras on data sets as large as 3 GB; projects also involved SQL, Spark, and web-scraping• Capstone project used a 12 GB data set to analyze sentiment of Twitch chat rooms; used VADER and Facebook's fastText and built a neural network (LSTM model) inspired by state of the art research on NLP to predict sentiment; currently developing app for end user that display daily sentiment of streaming channels for specific games• Practiced python coding challenges on algorithms and data structures daily

University Park, Pa, Us

• Collaborated on an algebraic geometry research project (listed under Research) to extract patterns from data and create a general formula that explains the dimensions of certain vector spaces; presented the derivation of this formula at four seminar talks and five conferences, including one major conference at the University of Illinois-Chicago• Instructed 300+ students in linear algebra, calculus, and ODE's/PDE's over the course of eight semesters• Served as a TA for an algebraic geometry course on polynomial interpolation with 11 students; supervised and judged research projects as well as educated students on the language and history of modern algebraic geometry• Organized a seven-month-long special topic reading group with over ten participants and eight different speakers, allfrom diverse backgrounds; maintained group webpage (HTML) with TeX'd notes

• Worked with three colleagues to input and evaluate client portfolio values and metrics to build 401k plan presentations• Expanded on communication skills and practiced persistence through conducting cold calls to obtain new business• Scoured news communications and evaluated quarterly reports to generate a stock analysis to present to team leaders up to five times per month

Doctor Of Philosophy - Phd, Mathematics

Bachelor Of Science - Bs, Mathematics

Education record