Ayushi Bansal Email and Phone Number

AI & AUTOMATION TEAM : *Leading the team of 10+ members. * Build the end-to-end POC for classifying emails and attachments into exceptional or good-to-go categories for IPFS client based on the needs they mentioned. Technologies used in the entire framework are the DONUT model, LLAMA2, flask API. * Build the end-to-end POC for searching the content from DB using Microsoft Azure Search Services. Technologies used in the entire POC are Azure AI cognitive Search, Vector search, Azure Storage, and Flask API. * Build the end-to-end POC for assisting the client in their tax-appealing documents and settlements. Technologies used are VectorAI - PALM2 model (text-bison), google embeddings, Google models, OpenAI models.

Product and Technology Team:Working as a core member for developing different modules using AI and ML techniques in video and content moderation's: (Working in a team of 2 people)*Develop the tagging module (tagging basically comprises various things and use-cases available for the particular video) using NLP techniques.*Develop the automatic promo and trailer creations using multiple AI-ML techniques.*Develop the tool for identifying and clustering unique persons present in the video using ML algorithms.*Develop various algorithms for removing background music and action recognition inside the video .*Develop the end to end Multi class Classification pipeline for classifying the Theatrical Cues of an videos. *Modify the existing subtitle player for star shows by implementing certain AWS skills.*Developing Video taggers, promos, teasers, and trailer generation for various clients using ML techniques.*Generate speech-to-text using Azure services.* Classification of Theatrical cues dataset using Pretrained and in-house models.*Modify OpenAI whisper as per our in-house dataset for generating subtitles with good accuracy.*Develop a search module for searching videos by using various NLP and ML techniques.

Medical and Life Science Group:*Working in a 5 member team.* Currently part of project "Chest rAI" - An AI-based chest X-Ray analysis system to assist Radiologists, which has won the "Big Innovation Award USA - 2021".* Co-inventor in Patent "202041014479 – System And Method OF Anomaly Detection AND X-Ray Report Generation".Contribution:*Design architecture for classification and detection of abnormalities on chest x-ray. Use the pre-trained model for comparison.* Design Localization Metrics for verifying Doctor vs Ml localization.* Localizing diseases onto Chest Xray using Object Detection algorithm like UNET, Mask R-CNN, Faster RCNN, etc.* Ensembling the classification models for improving localization output.* Generate the text classifier for medical report classification using NLP, on top of the existing Chexpert Labeller.* Report generation using a transformer and LSTM algorithms.* Working on other medical use-cases like knee osteoarthritides, coronary use cases like stenosis, calcium scoring and plaque characterization. * Train junior and freshers for the current working task and knowledge.* Collaborate with IIT-Bombay and other organisations regarding model and accuracy improvement.

Detecting various anomalies on chest X-Ray, refineries, powerplant, and aeronautics images- Preprocessing of images using AI-ML algorithms - Use various pre-trained CNN networks to train the model.- Work on creating the new architecture from scratch using multiple convolution layers.- Fine-tune various parameters, used different skills and methods for accuracy improvement.- Designed the Natural Language Processing Labeler for text classification.

Research Intern:* Worked on a real-time project with the Indian Government dataset under the guidance of the head of the Photogrammetry Department (ISRO, DEHRADUN).* Analyse the forest fire dataset collected with the help of the satellite, perform EDA on the dataset with the help of various Machine Learning algorithms and computer vision techniques.* Classify the dataset into the fire and non-fire classes with the help of Deep Neural Networks. * Used Keras and Tensorflow frameworks and python coding languages.