Ashyam Zubair

Client: Dubai Police (Jan 2020 - Oct 2020)
Project: Futuristic Services
- Digitize major policing operations using Artificial Intelligence and saved several man-hours and efforts.
- Performed EDA on the data provided from multiple sources and identified key features that can be used for the development of the solution.
- Provided visual explanation of the data and performance of the models developed and deployed.
- Applied Natural Language Processing techniques to process Arabic text data from historic crime data to classify crimes using ensemble models.
- Developed solution to automatically generate a consolidated case summary based on data collected by the Officers and applying NLP techniques & own algorithms to finally aid in downstream process.
- Mitigated heavy imbalance in data distribution using Synthetic Minority Oversampling Technique.
- Developed APIs for modules using Flask to assist UI team to integrate with their in-house proprietary application used by the end users, and display the results of the AI system.
- Containerizing the entire solution using Docker for seamless integration and migration between development and production environment.

Client: Franklin Templeton (August 2018 - December 2018)
Project: Automation based on Letter of Instruction
- Developed OCR solution to read instructions sent by the users in free text through emails, PDF files, scanned letters and bank cheques attached along with it.
- Developed algorithms to extract key information from bank cheques.
- Developed signature detection module to detect signatures from the cheques.
- Trained entity recognizer using Rasa NLU to extract key features from the letter of instructions sent by the user.
- Developed intent classification models to recognize the intent of the letter and perform actions based on the entities captured from it.
- Coordinated with the UI team to integrate the solution and develop an intuitive way of training new letters of instruction.