Lisa Deng Email & Phone Number

Mountain View, Ca, Us

Redmond, Washington, Us

Software Engineer currently working on Azure Sentinel. Providing platform support and enriching security data to provide a broader context and better hunting experience for security analysts in Azure Sentinel.

Redmond, Washington, Us

I worked in the Cloud AI Security organization within the SIPS ML Detection Team. The SIPS team focuses on building machine learning based cross service detections. I worked on a requested feature that involved building a machine learning classifier to classify certain kinds of user accounts. The classifier was a requested feature that would aid both my team and another external team's detections.During my internship I:- Built and trained a RNN for account classification using CNTK python (Microsoft Cognitive Toolkit Library) that achieved 95% accuracy- Worked with other teams for model improvement.- Researched deployment options such as CNTK C# and Azure Machine Learning.- Deployed my model into production using Azure Machine Learning.- Created a pipeline for future model deployments.- Presented my findings to my organization.

Boston, Massachusetts, Us

A majority of scientists at Massachusetts General Hospital (MGH) spend time identifying relevant patients for case studies. One source of data is written doctors notes on patients, identifying prognosis, patient characteristics, and other details. Our long term goal is to train models to classify these doctors notes into categories pertaining to medical topics and patient types. However, patient data is sensitive information. In order to publish results and run experiments, we will use Pubmed article abstracts as a surrogate for doctors notes. Most articles on Pubmed have an abstract along with a major MESH term (Medical Subject Headings). We can use the abstracts as a corpus, and the MESH term as a label. I am currently training machine learning models such as Logistic Regression and Naive Bayes for MESH identification using the Spark library.

Tel Aviv, Il

A common problem space in many growing companies is the need for internal service monitoring. At Kaymera I did research on popular microservice stacks such as the ELK stack and Graylog. In the 5 months I spent there, I:- Set up the ELK stack- Set up pipelines for information retrieval- Gained experience with Docker- Created custom alerts for problems using ElastAlert- Used Curator for log rotation- Gained experience using Linux- Contributed and learned from Elasticsearch discussions- Researched heavily into and used Kibana and Logstash plugins for other customized needs- Wrote documentation - Pushed the microservice monitoring stack into production

New York, Ny, Us

Designed graphs on Kibana that provided user activity insight for the sales teamCreated a Django Admin UI for analysts to manage dataBuilt data visualization graphs using d3.jsSet up a RESTful API with Django Rest Framework to plug into other apps

Cambridge, Ma, Us

Database management, updating web caches, writing regression tests, and learning more about the complex systems at the MIT Infolab.

Working on building a complex parsing algorithm in Common Lisp to load entries into a dictionary.

An Etsy business that specializes in selling hand-crafted origami pieces. We specialize in bouquets, flowers, and gifts for special occasions such as weddings and anniversaries.

Santa Barbara, Ca, Us

Participated as a venture founder in the Catapult Incubator for young entrepreneurs. Brought my company Everything Origami to the next level through iterative feedback testing and selling my products at boutiques in Harvard Square.

Bethesda, Md, Us

Worked as a summer student at the National Institutes of Dental and Craniofacial Health by conductingbioinformatics analysis and analyzing data.Published as Co-author in a Nature Immunology article titled: The DNA-binding inhibitor Id3 regulates IL-9 production in CD4+ T cellshttp://www.nature.com/ni/journal/v16/n10/abs/ni.3252.html

Bachelor’S Degree, Electrical Engineering And Computer Science

Education record