About
A team leader with experience in moving forward the data ecosystem, getting data science projects off the ground, organizational management and bridging gap between technology and business strategy. I have a proven track record of enabling high performance teams and driving new data science products from idea generation stage to implementation.
Some career highlights that demonstrate examples of being a change agent, persuader and effectively dealing with high level executives are:
1) Invented and deployed in production a new machine learning based campaign performance report. This enabled Yahoo to measure the causal relationship between advertising and customer action, without doing A/B testing. The project let to a patent, 3 peer reviewed publications and incremental revenue for Yahoo.
2) The Weather Channel: Invented and deployed in production a novel machine learning algorithm that enabled estimation of long-term avails.
3) The Weather Channel: Deployed in production a novel advertising campaign performance report that enabled holistic assessment of campaign performance over the whole network, using a variety of metrics.
4) JPMorgan Chase: Invented and developed a novel, machine learning based, database security system. This enabled simultaneous screening for intruders across all company databases.
5) Impacted budgets, collaboration and culture.
I have 20 years of building machine learning models and bringing them into production. Projects have spanned various domains including cybersecurity, advertising technology, finance, computer vision and neuroscience.
I hold a BA from Columbia University and I wrote my Ph.D. thesis on machine learning at Mount Sinai Medical School.
DATA SCIENCE:
15 years of experience in data mining, predictive analytics, machine learning and multivariate statistics.
LANGUAGES & SOFTWARE:
Python, R, Matlab, SQL, Hive, Pig, Hadoop, Spark, Netezza, Unix shells, Bash Scripts, Java (1 yr.), C/C++ (1 yr.), Excel and Photoshop
Ask me about
Work experience
- November 2018 – present
Northwell Health
Head of Data Science and Engineering ( Hiring)• Oversee hiring, budget planning, data science roadmap and technical strategy design, architecture and model selection for a team of data scientists and data engineers. • Drive innovation in the areas of patient experience, hospital throughput, advanced illness, patient digital experience, and patient monitoring. Developed new products in these areas and brought them from idea generation all the way to enterprise wide production. • Establish strategic goals of the IT department to ensure alignment with the goals of the business. • Planning of data identification, storage, provisioning, process and governance. • Responsible for strategic planning and execution of data strategy. Publications: • Lily Yung, et al. and Marsha Meytlis INTEGRATION OF A PREDICTIVE MODELING OF PATIENT EXPERIENCE TO PROACTIVELY IMPROVE CARE. Abstract published at SHM Converge 2022. Abstract OP7 Journal of Hospital Medicine. • Bari V, Hirsch JS, Narvaez J, Sardinia R, Bock KR, Oppenheim MI, Meytlis M. An approach to predicting patient experience through machine learning and social network analysis. J Am Med Inform Assoc. 2020 Dec 9;27(12):1834-1843. • Tóth V, Meytlis M, Barnaby DP, Bock KR, Oppenheim MI, Al-Abed Y, McGinn T, Davidson KW, Becker LB, Hirsch JS, Zanos TP. Let Sleeping Patients Lie, avoiding unnecessary overnight vitals monitoring using a clinically based deep-learning model. NPJ Digit Med. 2020 Nov 13;3(1):149. • Bari V, Hirsch JS, Narvaez J, Sardinia R, Bock KR, Oppenheim MI, Meytlis M. Using machine learning to predict the influence of physician communication on patient experience AMIA 2020 Virtual Annual Symposium Poster · Nov 1, 2020. • Coppa K, Meytlis M, Hirsch J, Horsky J. Automated Prediction of Patient Stability and Discharge within 36 Hours. AMIA 2019 Annual Symposium · Nov 21, 2019 • Brar R, Hirsch J, Meytlis M. Automated Identification of Patients with Advanced Illness Automated Identification of Patients with Advanced Illness, R/Medicine 2019 · Sep 12, 2019. - July 2016 – July 2018
JPMorgan Chase & Co.
Data Science Lead• Design, develop and deploy data science models to solve cybersecurity problems. Work with supervised and unsupervised machine learning algorithms such as nearest neighbor anomaly detection, regression, deep learning, random forest and principal component analysis. Use Python, SQL, Matlab, and Spark in PySpark environment. • Solve problems in areas of database security, application security, cloud security and email security. Collaborate on projects in data warehousing design and implementation, data architecture, and ETL. • Partner with Asset Management and Investment Banking teams to identify external and internal security threats in databases and applications. • Present R&D products and solutions to internal and external audiences. Developed a cutting-edge solution to the database security problem by detecting anomalies in database logs. - December 2014 – December 2015
The Weather Channel
Director of Data Science, Team Lead• Lead a team of data analysts and data scientists. • Created the data science road map to drive improvement in consumer advertising. Lead both operational and research data science projects to drive revenue for the company. Oversaw hiring, architecture and model design. • Performed QA and troubleshooting of models created by team members. Worked with models such as regression, and random forest. • Developed a new model that significantly increased prediction accuracy of ad inventory. Worked on projects related to market basket analysis, ad performance reporting, online recommendation engine for weather targeted ads, long-term weather predictions and personalization. • Researched vendors and evaluated third party data. - May 2013 – October 2014
Yahoo!
Applied Scientist• Oversaw analytics research projects of 3 data scientists. • Lead projects from idea stage to technical deployment. Communicated results back to sales and senior leaders. • Applied machine learning techniques such as regression, support vector machines, decision trees, collaborative filtering, and propensity score matching to data such as clicks, conversions, site visitations, online search, credit card purchases and demographic data. • Developed a new model-based approach to measure uplift, or true ad effectiveness. The novel causal inference-based algorithm was patented by Yahoo and led to 3 peer-reviewed publications. The results were rolled out in a new revenue-generating, client-facing report. • Developed a new approach, called Synergy, to accurately measure the value of brand awareness campaigns. The results were rolled out in a new revenue-generating, client-facing report. • Developed statistical algorithms for consumer advertising and personalization. Specific projects focused on the purchase funnel, online user journey, ad rank replay engine, multi-touch attribution, ad pricing, matching of offline and online data, and quality reporting. • Lead R&D of audience building algorithms at Genome/Yahoo. Results of research have been presented at Yahoo 2013 Summer Science Week and 2013 Yahoo Tech Pulse conference. - January 2011 – January 2013
Yahoo!
Senior Data Insights Analyst• Apply machine learning techniques to online data containing billions of points and thousands of features. • Develop statistical algorithms to predict which consumers are most likely to respond to specific online ads. • Developed a new method for objectively comparing the performance of behavioral targeting models. Presented this research at the YAHOO 2012 Summer Science Week. • Work on a new approach to optimize ad ranking in order to maximize network performance and revenue. • Build behavioral targeting models and optimize online advertising campaigns by finding the best audiences. • Trained new data analysts and oversaw their projects. PRESENTATION(S) YAHOO 2012 Summer Science Week: Model Competitiveness in the Real World - January 2007 – October 2009
Weill Cornell Medical College
Post-Doctorate Research Associate in Data Mining• Used generalized linear models and gradient descent optimization to solve problems in computational neuroscience. Developed Bayesian models of sensory information processing by the brain. Studied how neurons encode information in the brain. • Used information theory to predict visual images seen by the eye from the electrophysiological signals recorded in the mouse eye. • Disproved a hypothesis that groups of neurons encode a significant amount of information in neuronal correlations. These results were presented at the annual Society for Neuroscience Conference in Chicago, 2009. • Managed 2 graduate students working on the project. PUBLICATION(S): Meytlis, M., Nickols, Z., and Nirenberg, S. Determining the role of correlated firing in large populations of neurons using white noise and natural scene stimuli. Vision Research (2012), v.70, pp. 44-53 Meytlis, M., Bomash, I., Pillow, J., and Nirenberg, S. Assesssing the importance of correlated firing using large populations of neurons, Society for Neuroscience (SFN) poster (2009) - July 2003 – July 2007
Mount Sinai Medical School
Graduate Researcher in Data Mining and Machine Learning• Advanced a range of classical and state-of-the-art models in computer vision. Worked with machine learning and statistical algorithms such as principal component analysis, singular value decomposition, linear discriminant analysis, support vector machines, regression, neural networks, Fourrier transforms and spatial frequency filtering. • Developed solutions to several classical image recognition problems. Performed analysis of images such as photographs of human faces, photographs of objects, and functional MRI brain images. • Designed an algorithm that enables computers to recognize photographs of human faces with 99% accuracy, despite variations in lighting. Potential applications include security, biometric authentication and automobile driver assistance. • Determined the minimum number of dimensions that are necessary to identify a face image by a computer. • Developed a new method of lie detection that enables a machine to determine if someone is familiar with a face. PUBLICATION(S) Marsha Meytlis and Lawrence Sirovich On the dimensionality of face space, Transactions on Pattern Analysis and Machine Intelligence (2007), v. 29(7), pp. 1262-1267 (featured article of the month) Lawrence Sirovich and Marsha Meytlis Symmetry, probability and recognition in face space, PNAS (2009). V. 106(17), pp. 6895-6899 Marsha Meytlis, A model of internal face space, Visual Cognition (2011) V. 19(1), pp. 13-26. Marsha Meytlis and Cheuk Tang Lie detectors for face recognition, Neuroreport (in preparation) - January 2001 – June 2002
SUNY Downstate Medical Center
Research Assistant· Worked on technology for interfacing the rat brain with a computer. Trained remote-controlled rats.
Education
- 2003 – 2007
Mount Sinai School of Medicine of New York University
Doctorate in Machine Learning, Thesis in machine and human face recognition - 2001 – 2003Candidate for Doctorate in Computational Neuroscience, Neuroscience
- 1997 – 2001BA in Biology from Columbia College