Shahar Soudry

• Developed and deployed an anomaly detection monitor using time series models to reduce detection time in quality control processes • Designed a dynamic root cause classifier model tool to enhance diagnostic coverage and problem-solving in manufacturing lines • Created a hybrid outlier detection method aligned with statistical distributions detector for improved accuracy • Scaled and integrated models using GCP services, applying algorithms such as LightGBM, ISO, LOF, DBSCAN, ARIMA and EMA.

• Developed and deployed AI solutions for the Ministry of Health, with predictive models that improved early detection and risk assessment of medical conditions, including eye diseases and pressure ulcers • Published 4 papers and led data engineering efforts, along with text analysis and LLM methods, to create medical data lakes that enable large-scale future research • Utilized ML algorithms such as CatBoost, XGBoost, GBM, BERT and T5; collaborated with an ML Ops engineer to implement data integration pipelines using AWS Glue, and supported large-scale querying and analytics via Amazon Redshift

• Built credit risk models for the US SMB lending fintech market, optimizing credit terms through yield prediction using credit bureau data (Experian, Equifax) and internal features • Utilized ML algorithms including CPH, Logistic Regression, Elastic Net; advanced data analysis and data engineering in scale using BigQuery

• Led diverse ML solutions in the green energy market, including survival modeling for defaults, product analysis, customers' churn prediction, users' clustering and time-series revenue forecasting • Applied advanced statistical methods: XGBoost, Random Forest, Weibull Distribution, KMeans, DBSCAN; mentoring of survey-based analyses to support data-driven decision-making

• Researched and developed marketing prediction models in the telecom sector, focusing on churn, retention, campaign effectiveness, offer timing, and next best offer strategies • Utilized ML and statistical methods: Logit, CART, hypothesis testing in a big data environment using Hive for scalable querying and data processing

• Led risk modeling projects for the Israeli banking sector, mainly liquidity risk and repayment prediction models for FIBI main products • Developed credit score models (PD, LGD, EAD) for Mizrachi-Tefahot; recognized as the company’s outstanding employee in 2013!

• Medical and economic analysis for the clinic decentralization department • Long term strategies plan for clinics’ medical services • Doctor’s profile development and economic efficiency examinations