Snehil Shukla

Data Scientist equipped with a B.Tech degree from IIT Kanpur with 1 year and 7 months experience working in a leading Insurance company.

Expert in handling Data handling, cleaning, interpretation and building Data driven solutions through AI/ML and Natural Language processing techniques.

Experience in automating systems through Image processing/OCR in tandem with the skills mentioned above.

Experienced in integrating Generative AI models like Chat-GPT for custom solutions using prompt engineering for better results

• Email Bot: Developed a model to identify intent and create a draft based on the respective pre-defined logic flow to help Customer Relationship agents. Modelling involved Natural language processing techniques and various Machine Learning and Deep Learning models. Also worked with Chat-GPT to check if higher accuracy can be achieved using generative AI, Deployed the solution using Azure functions
• Customer Data De-duplication: Worked with a small team to develop an in-house de-duplication pipeline on Azure to cluster duplicate entries into the same group so that data quality for customer level analysis is improved.
• Claims Anomaly Detection: Worked with a small team of 3 members to explore methods of detecting anomalies in Claims made by customers. Explored the generated data extensively using Tableau to possibly identify any criteria for anomaly detection . Built an isolation Forest model and evaluated it for possible use in anomaly detection
• Stroke Prediction: Developed a Machine learning model to predict likelihood of a customer having a stroke so that steps can be taken to reduce company losses by identifying those who are vulnerable.
• Home Claims Project: Learned image processing and Object classification model development through Deep learning
  and Bounding Box based approach . Worked on optimizing the code and deployment for a project to determine the total damaged area
  calculation
• Health Product recommendation: Built an API using Azure Function App that would provide a product recommendation using available data through company's website. A cohort based approach was developed using customer data to identify the suitable product