Aayush Jain

Led a team focused on advanced machine learning and predictive modeling for signal processing applications. Spearheaded the development of predictive models using Recurrent Neural Networks (RNNs) and Spectral RNNs. Worked on quantization techniques to optimize model performance in resource-constrained environments, resulting in efficient deployment on FPGA platforms. Collaborated with cross-functional teams to integrate FPGA-accelerated models into real-time data processing pipelines, enabling rapid decision-making. Developed secure Deep Authenticator for vending machines using MTCNN and FaceNet embeddings, ensuring seamless and safe customer transactions. Streamlined communication of technical concepts through documentation and presentations for diverse stakeholders.

Successfully built a domain-specific, conversational and rule-based chatbot tailored to client specifications, contributing to improved user interactions and client satisfaction. Implemented advanced NLP techniques, including domain and intent classification, entity recognition and resolution, and language parsing, leveraging LSTM and BiLSTM models for accurate entity detection. Contributed to the development and testing of APIs, ensuring seamless integration and functionality of the conversational application within the existing infrastructure. Engaged in quality assurance and testing phases, validating the functionality, robustness, and accuracy of the conversational application across different scenarios. Played a key role in preparing and labelling invoices, significantly contributing to the training dataset for Invoice Automation Using OCR, a critical component for enhancing automation accuracy.

Successfully executed a Silicon Wafer Defects Detection project by leveraging advanced Machine Learning techniques. Devised a novel deep learning architecture that integrated Convolutional Neural Networks (CNNs) and transfer learning techniques, enabling the accurate identification and classification of intricate defects. Collaborated closely with domain experts and engineers to fine-tune the model's performance, iteratively improving accuracy and reducing false positives.

Built NLP, DL and opencv based application for user and Performed image classification using resnet, VGG16, Inception networks and used Object Detection Algorithm such as yolov3, RCNN to detect objects. Assisted in the collection, preprocessing, and cleaning of raw data from various sources. Developed python scripts for data manipulation for machine learning and perform Preprocessing of data using pandas, numpy, matplotlib and seaborn.