Rahul Reddy Vemireddy

• Have built - Pipeline to include OCR for Text Parsing using Tesseract, an open source deep learning based OCR Tool to enable text based search on the attachments of outlook emails.
• Built Alerts system for Supervision, a product that scans through every email/ teams chat and flags if it's offensive language or bad words.
• While working on family safety application: - from scratch, built a VPN on iOS to monitor network traffic and application usage - worked on improving the entire onboarding of a user.
• This is critical as it touches the core of the application Lead a team of 5 people to build the VPN and 5 people to improve the onboarding experience.
• Have lead a v-team of 5 devs and setup processes to improve testability, accessibility and localization for the mobile applications and services
• Currently working building customer insights for the entire application and the applications using family platform. This system would then handle a scale of 15PB every day.

Oversaw entire product deployments from vision and creation to rollout and delivery.

• Software Engineering - Developed an end to end text classification and sentiment analysis system.
• Lead a team to Develop a platform that processes ~2Bn Reviews daily gathered from social media sites
• Data Science - Developed applications using both machine learning and deep learning techniques to classify the social media data relevant to a given industry or domain.
• Developed a theme creation process based on LDA and Pos Tagger.
• This reduced the turnaround time of training data creation by 70%.
• Developed classifiers for text classification and sentiment analysis, for state of art results.
• By a combination of a few techniques, the current 2-point sentiment analysis process works at ~85% accuracy.
• Web Development (Ops flow) - Developed a workflow to QC topics & sentiment classification and generate training data.
• This system handles versioning for several iterations between analyst and the user.
• The analysts create categories using this workflow with the help of category suggestive keywords and keyword pairs produced by the workflow tool.
• Implemented user hierarchy, where admins can overlook the users and user created jobs/projects.
• Using React, d3.js, several dashboards to view insights generated by the pipeline.

Revised, modularized and updated old code bases to modern development standards, reducing operating costs and improving functionality.

• Using NLP and machine leaning techniques, I have built a system to automate text answer correction for mega exams.
• Developed several web portals and APIs for customer companies and an in house product, mega exams.