* Develop proprietary AI algorithms for trading with Interactive Brokers.* Over 100 YouTube videos in Python, ML, and algotradinghttps://www.youtube.com/@MLAlgoTrader

* Predictive Maintenance: Predicted Failures in large AI DGX GPU LLM Clusters.* RL for Video Compression: Used PPO + SAC to find optimal compressed encoding of videos* Cable Failure Prediction: Predicted Failures in Ethernet Cables along the testing stages before reaching production.

* Financial Research Distribution: Created a system for matching JPM clients to relevant research documents in JPMs vast financial research ecosystem based on relevant interests.* Fraud Detection of Wholesale Payments: Created fraud detection model for detecting fraud as well as reducing false positive rates of existing rule engines alerting on unusual activity. Anomaly Detection of Data Feeds: Found anomalies in data feeds (i.e. tick-latency and frequency)* Spark for Data Scientists: Taught a weeklong seminar for data scientists across JPM worldwide.Deep Learning in Tensorflow: * Gave a seminar on Deep Learning and Tensorflow for Israel Office.

* Lifetime Value Prediction: Created an end to end framework and model for estimating the future net margin, cost, and net contribution a customer will provide to Agoda in the next 3,6,9,12 months..* Clustering of Hotels: Found groups of hotels in which users are indifferent between.

Develop Machine Learning Systems and Statistical Models for verification of online ad impressions to insure they are viewed by human traffic as opposed to automated invalid traffic (i.e. fraud bots, non-human data centers) and perform POCs on terabytes data. Projects include: ● Site Similarity and Clustering based on User Visits: cluster similar sites by user visits to find new sites with fraudulent activity as well as clusters of sites visited by bot networks. ● Emulator Detection: detect ad impressions coming from emulators as opposed to physical mobile devices based on abnormal device sensor data such as accelerometer values. ● Fraud Bot Detection: create new fraud detection algorithms for finding ad impressions with some form of invalid traffic based on traffic anomalies and enhance current fraud detection methods by finding users with similar behavior to known fraudulent users.

Provide leading customer experience analytics for giant telecom providers. Projects Included: ● Churn Analytics: Predict the likelihood of customers canceling their subscriptions based on their quality of experience and other relevant factors. ● Proactive Analysis: ​ Develop algorithms to detect problems in the network in real time. This is done with an ​intelligent thresholding algorithm, which learns a “normal” pattern of the data and alerts as soon as a sequence of observations (i.e. # of dropped calls in an interval) is anomalous. ● Trouble Ticket Resolution Predictor: ​ Create an automated trouble ticket resolution classifier: for each trouble ticket I calculate a customer’s KPIs up until the time the ticket is created and predict how the technician will resolve the ticket. ● Device Performance Dashboards: ​ Create dashboards, drill downs, and heat maps in Tableau for analyzing device performance for successful POC leading to new product. ● Sentiment Analysis:​ Use Twitter feeds to analyze sentiment of subscribers, service providers, and different device types. ● Data Mining Massive Sized Telecom Records: Find interesting patterns among subscribers and devices, and other insights from tables with up to 80 billion records.