Risha Raj

• Developed and implemented a credit card behavior score model for Marks & Spencer (M&S) Bank,
  leveraging machine learning techniques to assess customer risk and optimize credit strategies.
• Utilized historical transaction data, payment patterns, and credit utilization metrics to enhance predictive
  accuracy.
• The model improved risk segmentation, enabling more precise credit limit adjustments, proactive fraud
  detection, and better delinquency management.
• Developed and documented the Secured Essential Expenditure Model for affordability assessment in retail
  mortgage applications of HSBC customers.
• Designed the model using neural network and ML to predict applicant expenditure across key categories.
• Conducted model testing, validation, and performance monitoring to ensure accuracy and regulatory
  compliance.
• Collaborated with cross-functional teams to integrate both models into decision-making systems, leading to
  improved portfolio performance and reduced default rates.

Cisco

• Automation using ML for Sales Compensation/ Reporting
  for Competitive Intelligence Team/GCP deployment.
• Creating BERT embeddings for text based data for the
  purpose of calculating similarities.
• Performed predictive analysis using machine learning
  models and presented dashboards to client based on the
  predicted results.
• Integrating ML model with a rule based model.
  Collating data from multiple sources, processing the data
  in PyCharm and Snowflake and creating Tableau
  dashboard.

AstraZeneca

• Data Extraction through NLP.
• Worked extensively on HRM, a Human Reading Mimicking
  module which mimics the way humans read and interpret
  an unstructured text.
• Utilised HRM to extract text based entities and their
  relationships from medical forms.
• Preprocessing using Spacy and NLTK.
• Built Supervised Machine Learning Models for
  classification.

• Performed an experimental study of classifying dialogue
  acts using 3 multi domain datasets.
• Classification using various machine learning algorithms.
• LinearSVC, Stochastic Gradiet Descent, Multinomial
  Naive Bayes, Decision Tree, Random Forest, Neural
  Network, Multilayer Perceptron.
• Performed count vectorization and Tf-Idf on the dataset.