Arvind Narayan Email & Phone Number

Atlanta, Ga, Us

New York, New York, Us

Using a variety of data science techniques to interpret and action on data relating to WMG's artists, songs, and fans.

Atlanta, Georgia, Us

Developed and productionized a machine learning pipeline to highlight Trial and Freemium accounts with a high probability of paid conversion. Created a highly accurate churn prediction score using techniques including imbalanced dataset splitting, feature importance ranking, and unsupervised learning. Worked with C-level executives from product, sales, and engineering teams to present, socialize, and iterate on these models. Led engineering and product teams in "how might we" exercises and created a Cox Regression model to establish FullStory's North Star metrics. Developed and implemented a data model for Looker, bringing together product data, Salesforce data, and accounting data. Built various algorithms for internal use cases, including URL canonicalization, modeling customer journeys, and web traffic analysis.

Atlanta, Ga, Us

Generated predictive models for revenue and hospital write-offs using data wrangling and machine learning techniques with SPSS, Python, and R. Accurately predicted hospital write-offs with 5% average error. Modeled bed throughput using a combination of individualized clinical data and overall performance metrics. Created reports and ETL processes in Netezza SQL and Microsoft SQL Server to analyze key clinical and financial data. Modeled patients' financial outcomes using an ensemble model with traditional top-down forecasting methods and bottom-up machine learning methods. Created Spotfire dashboards to streamline accounting processes and to display predictive models.

Cumming, Georgia, Us

Linked dealership retention rates to brand and demographic information using supervised and unsupervised learning techniques. Performed win/loss analysis using a targeted customer survey to identify market expansion opportunities of $600K ARR. Extracted data from Salesforce and integrated R with QlikView to highlight key correlations between dealers' locations and marketing needs. Used Google Analytics to compare KPIs of Client Command's digital marketing tools with competitors' tools, correlated those with business performance, and identified Client Command's key competitive success factors.

Mountain View, Ca, Us

Used sensor data from mobile devices and machine learning techniques to infer user activity and display relevant information. Drew from a data set with millions of individual data traces and preprocessed raw sensor data to extract meaningful handcrafted features. Added additional features to simulate real-world user behavior. Executed an ablation study to determine which features had the largest impact on classification performance. Used Weka's Java API to automate model training, cross validation, and testing. Improved model classification accuracy by 27%.

Atlanta, Georgia , Us

Worked with Prof. Vijay Vazirani and Dr. Ruta Mehta to model supply-demand interactions in real-world markets. Modeled these interactions using the linear complementarity problem, a well-known quadratic program. Used a variation of the simplex method for linear programs to solve the linear complementarity problem. Developed a MATLAB script to simulate a market environment with buyers and sellers and to achieve equilibrium using our derived method. Wrote data scrubbing algorithms and used existing MATLAB libraries to make the script efficient for large and sparse matrix parameters. Applications of this research can help companies optimize their market operations to increase profits.

Cincinnati, Ohio, Us

Worked at World Shaving Headquarters, Boston as a Process Engineering Intern in Blades and Razors. Wrote MATLAB and PBASIC code to improve the operations of an assembly station. Established an inventory monitoring system in a production machine lab and set up a plan for widespread implementation of the system, which will generate $700,000 in cost savings. Used 6S methodology to improve the safety and efficiency of manufacturing equipment.

Cincinnati, Ohio, Us

Redesigned a defective assembly station for an upcoming product using CAD and a root-cause analysis process. Designed and implemented robust experiments to determine the various effects of sharpening wheel attributes on grinding force. Tested and oversaw the implementation of a new tooling alignment assembly onto a production machine.

Designed tutoring curricula for the Mathematics, Writing, and Critical Reading sections of the SAT. Tutored students on test-taking techniques and subject proficiency.

Atlanta, Georgia, Us

Used rapid prototyping to create a power supply PCB that supported three other circuit boards with varying functionalities. Modeled the board schematic with PCB Artist, created a parts list and ordered relevant parts, and integrated these components using soldering techniques.

Bachelor Of Science - Bs, Computer Science

Master Of Science - Ms, Computer Science