Ayush J. Email & Phone Number

Richardson, Texas, Us

San Francisco, California, Us

• Built the company's first widely-adopted machine learning model for Churn Likelihood Prediction (82% accuracy and explainable).Helping significantly in the economic downturn environment to retain customers as early warning system.• Developed an Ideal Customer Profile (ICP) model used by sales and marketing for targeted segmentation, instrumental for ABX strategy.• As full-stack role, architected and built scaled DBT orchestration on Airflow, optimized data volume saving $83k in Redshift costs.• As full-stack role, architected and optimized CircleCI workflow, reducing build time by 90% and added Continuous Deployment feature.• Built new and maintained data pipelines on Stitch, Glue, DBT, and Databricks, processing Terabytes of data daily.

San Francisco, Ca, Us

• Built Seq2Seq Deep Learning model for “Predicted Next Role” in a candidate’s career path, increasing search results ~1.5x and relevance.• Built Machine Learning model to quantify candidate’s “Skill Proficiency” in different reported skills to improve search results by skills relevance.• Worked on “Envoy” product Model that surface best fit matched candidates for job criteria using GAME, GLM, with feedback learning.• Worked on improving and maintain the “More Likely To Move” model which predicts the candidates who are likely to change their job in the future.• Generated Email Insights with statistical significance for best practices in recruiting. Added 10s of “Entelo Insights” and badges for the profiles.• Built the contextual spelling correction model for search query using a statistical language model and Seq2Seq, to improve entity classification.• Built new & simplified old ETL pipelines for Big Data processing in Databricks and ingest to Elastic Search.• Maintain Data Science microservices.• Implemented numerous comprehensive analysis on 500 million profiles and datasets for actionable business insights.• Analysis on customer churn, search & all product features, user engagements, QBRs, KPIs, ROI and other metrics as business requirements then presented to Product Managers, Marketing, Sales and Customer Success teams.

San Francisco, Ca, Us

• Big Data Analysis on profiles with hundreds of feature columns – Performed extensive analysis for the direct Actionable insights, quality, outliers, and data validity using Spark in Databricks eco-system. Wrote complex SQL queries for aggregation and code for visualization.• Feature Extraction for Healthy, Spam or Sparse Profiles – Derived generic rules, features and validated for profiles classification Model.• Named Entity Linkage and Wikification for Entity Resolution.

New Hope, Pennsylvania, Us

Predictable Sales Revenue Tracking Model:• Developed the Model to track present sales and predict the future sales requirements and optimizations for targeted revenues.Work on Data for SEO and Email Marketing Analysis:• Analyze the market reach of the company and unifying multiple channels.• Helped with Database Management, cleaning data and acquiring more data for Marketing Lists. Automate the PvSyst Simulations:• Wrote a java application to Simulate the PvSyst software by modifying 30 variables in background.• Performs 50 simulations in an hour, saving time for a human to alter variables manually.• Get all the outputs by converting the image to text by using Google's Vision API.Earlier worked on proof of concept to Implement ML Recommendation Model for product users based on valuable insights from user interactions and history data.

New Hope, Pennsylvania, Us

Sales lead generation & forecasting Model:• Extracted Features and Patterns of product sales using Data Mining from Company Sales Data. • Engineered a Raspberry Pi cluster for Database Modelling and using Google API resources for knowledge extraction.• Performed NLP and Quantitative Analysis for features extraction on a database of 185K Construction Companies in the USA. • Created a predictive model using Machine Learning algorithms with efficiency: Logistic Regression(69%), SVM(84%), Random Forest(86%).Total Sales Representative Cost:• Performed an intensive market research of the Total Addressable Market for the Product on multiple databases of construction companies.• Used Google Geocoding for precisely locating 16.5K Target Companies within the USA.• Designed the algorithm to calculate the maximum targets for a Sales person from a given zip code.• Included location density, mode of transportations to minimize the spending and provide the best positions with maximum reach.

Us

Provide callers with excellent technical support and troubleshooting regarding various computer issues such as:- Assisting user with password troubleshooting or renew password Online or through Activation Key Support via Open Registry Authentication on phone. -Assisting users with RUWireless services, setting up game consoles to network via MAC bypass authentication on WPA2-Enterprise and 802.1X configuration.- Assisting user with Online basic virus troubleshooting using support software such as RADS.-Instructing callers on how to gain access to Rutgers Services, Net-ID related accounts and activate new services.- Configuring devices like iPad, iPhone, and other specific device manually for the Rutgers Email Services like RCI, Scarletmail, etc.-Assisting Rutgers Staff and co-Workers with issues. -Informing callers of scheduling/registration updates, transportation service changes, and other general campus-related questions.-Creating detailed and accurate support tickets for each caller. Also looking up the caller's account and service information in the System and Documenting all the changes or updates in System.-Participating in ongoing training to learn more about the systems and software and keeping up-to-date via e-mail, NBCS web sites and other resources

Bachelor Of Science - Bs, Computer Science

Summer Classes