Abhay Agarwal

Custom Module for Annotation

• Specify custom rules in pre-defined format in excel sheet
• Processing text using above rules for generating annotation
• Using this data to build classification model

Knowledge Extraction

• Ingested legal contracts and extract out text data to perform data mining
• Vectorized sentences using tf idf and doc2vec techniques
• Built RNN deep learning classification models to annotate sentences
• Achieved more than 90% accuracy on multi-label classification models.

Sports Analytics Mobile App

• Used BlazePose’s pose detector model for holistic pose detection
• Built classification model (with 95%) for activity classification
• Integrated full pipeline into android/iOS mobile app

Face Similarity

• Used MTCNN algorithm to extract faces from pictures of varying poses, expressions
• Created facial feature-set using VGG16 deep-learning architecture
• Milliseconds performance with accuracy comparable to existing algorithms

IoT Data Analytics

• Implemented complete pipeline for daily data analysis, deployed on heroku/docker
• Saved devices from failure and critical damages by generating signals
• Reduced maintenance cost by 20% by timely alerting

Loan Default Prediction

• Processed million records using PySpark with few seconds latency
• Got 92% accuracy with random forest model pipeline deployed on Heroku.

Web Applications

• Blogging App :- Implemented Medium like functional blogging web application using Django
• Social Media App :- Used MVT architecture of Django, created Facebook like application

• Maintain library of model documents, templates,& other reusable knowledge assets
• Gather business requirements from digital innovation teams of organization, draft analytical solution and Implementation
• Create prototypes of business problems using different algorithms classification, time-series, regression etc.

Customer Segmentation

• Analyzed customer spending data to send targeted promotions
• Improved conversion rate of offers uptake by 5%

Sales Forecasting

• Forecasted sales figures to help in production planing
• Reduced inventory cost by 15% and improved profits by 10%

Sentiment Analysis

• Mark twitter data for inappropriate content
• Worked with election commission to stop hate content on twitter.

• Synthesized current business intelligence or trend data to support recommendations for action.
• Analyzed technology trends to identify markets for future product development or to improve sales of existing products.
• Identified and analyzed industry or geographic trends with business strategy implications.
• Manage timely ow of business intelligence information to users.
• Communicate with customers, competitors, suppliers, professional organizations, or others to stay abreast of industry or business trends.