Aditya Jagdev Email & Phone Number

Bangalore Urban, Karnataka, India

Singapore

Bengaluru, Karnataka, India

● Led a team to create a framework for automating data narratives, summarize data behavior, and output findings in natural language. Used multiple model explainability techniques (statistical analysis, LIME, SHAP) to create a novel, human-understandable method of providing explanations. Wrote a white paper for the solution.● Developed a supplier search tool to allow clients to source goods from a wide range of suppliers. Augmented data using web scraping. Used roBERTa and Google USE to… Show more ● Led a team to create a framework for automating data narratives, summarize data behavior, and output findings in natural language. Used multiple model explainability techniques (statistical analysis, LIME, SHAP) to create a novel, human-understandable method of providing explanations. Wrote a white paper for the solution.● Developed a supplier search tool to allow clients to source goods from a wide range of suppliers. Augmented data using web scraping. Used roBERTa and Google USE to understand the context. Achieved 95% top-10 accuracy. Show less

● Developed and pushed to production a discount/rebate classifier that leveraged text clauses and numerical data related to discounts in scanned contract documents. Increased accuracy over previous implementation from 72% to 94%.● Contributed towards building an item description classifier with >150,000 labels. Integrated third-party data to create language context for the item description. Improved top-3 unsupervised classification accuracy from 62% to 87%.● Led a team to create a… Show more ● Developed and pushed to production a discount/rebate classifier that leveraged text clauses and numerical data related to discounts in scanned contract documents. Increased accuracy over previous implementation from 72% to 94%.● Contributed towards building an item description classifier with >150,000 labels. Integrated third-party data to create language context for the item description. Improved top-3 unsupervised classification accuracy from 62% to 87%.● Led a team to create a framework for privacy-centric, GDPR-compliant AI applications. Developed a federated learning pipeline and iterated through homomorphic encryption and differential privacy.● Created ETL pipelines for postgreSQL and ElasticSearch servers. Provided value to other teams by minimizing time spent on feature selection. Pipeline resistant to network failure and 350x faster than the existing implementation. Show less

Gurugram, Haryana, India

● Worked on distributed model training using Horovod over Azure ML Studio.● Studied map-reduce and parquet data format. Implemented regression scripts that used terabytes of data using PySpark

Gurgaon, India

● Created a tool leveraging sentiment analysis to detect unreliable reviews by comparing sentiment and given ratings.● Helped create online presence for clients using AngularJS for frontend and .Net Core for backend.

Gurgaon

● Created a solution for collecting in-person interview data.● Used statistical modeling to detect false/fabricated interview responses input by the interviewer (to prevent response spoofing).● Deployed the solution using Azure.

Gurgaon, India

● Focused on android app development.● Developed 'Right on Q' over a four week period.● Features include batch app uninstalling, recycle bin for apps with an added timer, and batch update applications on demand.