Aleena K Raj

Product Owner and AI Subject Matter Expert, responsible for Leading the Data and Learning Design team

Created a document parser using Object character recognition, Image processing and in-house boundary detection method to convert Textbooks to Digital textbook format for Content Taxonomy mapping. Developed BERT-based model to tag multimedia Content (youtube, WebM, pdf, HTML, ECML) in multiple languages. Tages were integrated into Content Search for better discoverability of 65k+ Content for 150M+ app users. Built and deployed Profanity Detection API using model based and Corpus based approach, to detect profane words in the content published by the author. Performance evaluation for the model conducted on manually tagged twitter corpus of 56000 records. Wrote an end-to-end Content enrichment pipeline that supports multimodal text extraction for contents of various mimetypes like audio, video, embedded URL, Images and PDFs in python by making use of Google cloud services like OCR, Vision and Speech-To-Text APIs. Cross-functional collaboration with the Data Engineering and Implementation team to automate detailed report generation for quiz consumption pattern for Users per session in Diksha Platform using pyspark

Worked in building ML-As-A-Service platform called ML-Workbench, an umbrella for tools and libraries in building and deploying Machine Learning Apps. Exploratory Analysis and development of Content-to-Vector pipeline for contents of various supported types like audio, video, images using ML algorithms like LSI and Doc2Vec.

Conducted Machine Learning Intensive Bootcamp for Wipro Elite Batch of 20 candidates. Wrote a script to implement Coreference Resolution on opensourced SQUAD Dataset using Stanford CoreNLP tool