Machine Learning / Optical Character Recognition (OCR) / Natural Language Processing (NLP) based application
Chatbot creation using AIML (Artificial Intelligence Markup Language)
Used React.Js to revamp UI
Worked on rule based engine in python
Created new hierarchy of linked tables for the application in SQL
Involved in model accuracy upliftment task
Worked with JUnits to increase code coverage
Wrote Unit test cases in python for repo
Provided production support for the application
Projects
-->Video to Audio Description • Project with an aim to convert video to text description and then to audio, which will be useful to people with visual impairments • Making use of LSTMs to convert video to text descriptions -->SIIM-ISIC Melanoma Classification (Kaggle) • Identification of melanoma in lesion images • Achieved a rank of 981 (top 30 %), with an AUC score of 0.9273 on final test dataset (highest score of the competition being 0.9490) • Used TFRecords, Tensorflow Dataset and Pipelines in Python language to increase the efficiency and speed up training • Overcame the challenge of heavily biased dataset, only 5 per of the total dataset was positive (having melanoma) • Worked with image processing techniques to preprocess the image data in order to achieve a better score Plant Pathology (Kaggle) Mar 2020 - May 2020 • Classification of plant diseases based on plant leaves images • Used Tensorflow, Keras and various image processing techniques in Python to achieve a 94.6% accuracy of classification • Handled image dataset with depth perception—angle, light, shade, physiological age of the leaf -->Celsius scale thermometer using 8051 Microcontroller • Made digital thermometer using 8051 Micro-Controller and Embedded C language • Scaled and fine-tuned the readings from sensors, to get an accurate reading on Celsius scale