Raju G

Experience

• 

  Perfectsoftskills

  Jun 2013 - Dec 2018

  Senior Manager

  ● Evaluated organizational training needs, collaborated with experts and senior leadership, designed solutions enhancingonboarding, performance, and quality management. Successfully improved compliance with PET manufacturing standards.● Maintained training materials database, conducted classes as needed to ensure ongoing learning within the organization.● Created training strategies, initiatives, materials. Managed organization including administration, banking, staff selection.

• 

  INNODATATICS * Innovation | Data | Analytics

  Jan 2019 - Apr 2022

  Machine Learning Engineer

  ● Analyzed data from 32 insurance providers to optimize employee health coverage, leveraging Python's Pandas for data processingand visualization, and K-means for clustering and recommendation engines, which generated $20,000 in company profits.● Developed and integrated a face recognition authentication system utilizing DeepFace for enhanced security and user experiencewithin an LMS application, which resulted in a 20% increase in company profits.● Developed a user emotion recognition model with dlib, integrating drowsiness detection via MediaPipe. Produced behavioranalytics reports for personalized engagement in LMS, which resulted in a 25% increase in company profits.● Developed an image-based mathematical expression extraction application utilizing Latex OCR, automating data entry forclients. This innovation generated $15,000 in profit for the company by streamlining mathematical expression processing tasks. Show less

• 

  OnPassive

  Apr 2022 - now

  Data Scientist

  ● Architected and developed a model to classify URLs as NSFW/SFW with 95% training accuracy and 94% F1-Score using GradientBoosting. Improved security for social media and email apps, saving $4500 monthly by reducing risks and enhancing user trust.● Architected and developed multi-class models with Gradient Boosting and XGBoost: one categorizes email content into fourclasses, enhancing organization and response efficiency; another classifies meeting agendas, resulting in a 30% revenue increase.● Architected and developed a binary classification model for NSFW/SFW text achieving 97% accuracy with XGBoost andDistilBERT. Enhanced content moderation, saving manual review time, and reducing moderation costs by $3500 monthly.● Developed Smart Replies model for email application using Dialogue GPT, trained with custom data. Contextually tailoredresponses enhance productivity, saving time. Contributed to 20% revenue increase by improving user engagement and retention.● Architected attachment classification for mailing and chat apps, using URL, text, and image models, which generate 25% profits.● Developed QA and MCQ generation, along with video summarization models using text-embedding-ada-002 and GPT-3.5-turbo. Show less