Robert Beatty Email & Phone Number

Atlanta, Georgia, Us

● Designed, implemented, and maintained the fraud prevention architecture. Transformed from a rules based engine within Socure to a model based approach built within Python using Scikit-Learn pipelines to score a model built using LGBM. Increased fraud capture by over 50% while decreasing fraud declines by ~25%.● Designed experiments for A/B testing onboarding process for new vendors as well as helped in designing the internal A/B testing system. ● Built pipelines and processes around regulatory compliance to manage 314a checks and used graph networks to check for fraud networks developing within the user base.● Maintained, designed, and implemented monitoring capabilities to track performance and to monitor our user base for early indicators of fraud/low performance so as to prevent future losses.● Utilized: Python, Docker, Poetry, AWS, FastAPI, Git and GitHub Actions in project development.

Atlanta, Ga, Us

● Productionized graph creation process for use by data scientist. Allows data scientist, via a config driven sparksubmit process, to use PII to create graphs which tell you how highly linked a piece of PII is to other PII.● Performed anomaly detection of suspicious log events within a customer facing product. Spark was used to perform the data engineering, Python and Dask via Yarn for the data science with the modeling done using XGBoost. Was able to increase detection rates 3 fold over the manual review process with minimal loss in suspicious events being detected.● Provided support to less technical members of the team. This involved: educating the team on tools like Spark, Dask, etc. Creating tools such as the above pii graph analysis tool. Performing technical/code reviews and helping team members use best practices such as using pyarrow to decrease data load times to python from the cluster by an order of magnitude.● Work is being done using: Spark, Dask, Python, Scala w/ SBT, and HDFS.

Atlanta, Georgia, Us

● Defined and implemented conversion of customer audit records of encryption key events generated by Ionic's ABAC policy engine from json to protobuf. That data was then converted to parquet and saved in s3 using a daily spark job.● Used the customer data to create numerous intermediate datasets that could be rolled up and used for doing customer analytics. Those intermediate datasets used sketches (hll, tdigest, etc.) and other streaming friendly data structures for merging the data easily.● Used the intermediate datasets to create analytics results saved as json, csv, and parquet that could then be used to construct reports or enrich other data sources. Analytics artifacts created: time series, tables with outlier metrics, bar charts, frequent pattern analysis, and anomaly detection.● Work was done using: Protobuf, Parquet, Scala w/ SBT, S3, Spark, and AWS Databricks w/ tooling written in Python.

Atlanta, Us

● Developed requirements for building out CNN Digital’s big data environment especially w.r.t. the analytics environment within AWS and supported the audience development team with bespoke analyses and statistical expertise.● Performed analytics on web traffic data using SQL, Python, LDA, and Graph Analysis. LDA was used to create weights connecting users which was then used as the edges connecting nodes for performing a community detection. Those communities were then analyzed for commonalities and trends.● Created several push alerts and bots for Slack within AWS using Kubernetes for conveying and obtaining CNN performance and social trending metrics from several different APIs.

Dallas, Tx, Us

● Performed a topic analysis using LDA on broadband users tech support notes to find common issues customers were experiencing which were then used to bolster the FAQ/Typical Problems part of the tech support website.● Analyzed historical data using random forests to predict the likelihood of a user returning to purchase accessories for a recent cell phone purchase. Those customers were then given discounts for accessories to entice them to purchase accessories.● Work was done using: R, Python, Pig, Hive, Teradata, and Aster.

Kennesaw, Georgia, Us

Instructed classes of 30+ in mathematics, developing and preparing assignments, quizzes, and tests to forward student’s knowledge level with a strong focus on problem solving and developing analytical skills.Courses Taught:• College Algebra• Trigonometry

Waltham, Ma, Us

Responsible for all the image processing, using IMAGIC-5, and data analysis using IMAGIC-5 and MATLAB of the STEM images in analyzing the structure of a polymer for possible use in target drug treatments.

Atlanta, Georgia , Us

Handled all of the lab procedures, sample preparation, and experimentation using a high speed camera and a florescence microscope as well as handling all of the image processing using MATLAB in the study of a connective tissue called hyaluronan's viscoelastic properties and how those properties change with the addition of binding proteins and changes in pH level and temperature.

Master Of Science (M.S.), Applied Statistics

B.S., Physics