Akshatha Ambekar

-Publication and presentation of scientific paper at WAISE (SAFECOMP) 2022. - Link to publication (Preprint): https://arxiv.org/abs/2206.11116 -Analysis and forecasting of multivariate time series for a predictive maintenance use case. -Support customers in development and integration of SafeML solutions into the ML production pipeline.

- Carried out several stages of AI product lifecycle from data preprocessing, designing algorithms to monitoring dependability of ML systems -Proposed and implemented effective novel reliability and robustness metrics for ML models considering statistical distance dissimilarity (SDD) between training and test environments. -Dynamic Time Warping based K-means clustering, anomaly detection for time series data. - Hypothesis testing, EWMA control chart, Deep Ensembles. - Python, R, PyTorch, scikit-learn

Time series forecasting using knowledge based Neural Networks - Surveyed healthcare datasets like pain medication, Inpatients, ECG5000 and Covid-19. - Univariate and Multivariate time series analysis and forecasting with R and Python. - Implemented statistical models and DNN models for healthcare datasets. - Hyperparameters search - Achieved performance improvement of 8.3% for Covid-19 data and 5% for Inpatients data contributing to state-of-the-art. - Python, R, Keras and Tensorflow

Handwritten Text Recognition using Transformers and Bidirectional LSTM - Implemented an architecture combining a Bidirectional LSTM with Transformer for short and long texts recognition. - Synthetic font data generation, pre-training, fine-tuning and dropout learning. - Data pre-processing and preparation, Tokenization, NLP, Hyperparameter optimization. - Achieved efficient text transcription on unseen text images with reduced model complexity. - Deep Learning frameworks: PyTorch and Fastai.

Constructed software architectures and designs for new features in safeTbox tool. Developed new functionalities for the tool. Enhanced usability and interoperability of the tool Tools and Technologies: C#, SQL and Mindfusion.

Software Developer : Web and Desktop Applications; Framework : .NET Responsibilities: - Designed and developed multiple software applications. - Integrated different modules of application to allow support for Linux and Windows OS. - Responsible for carrying out all stages of software development lifecycle. - Well versed with protocol buffers and messaging broker software RabbitMQ. Tools and Technologies: C#, C++, HTML5, CSS3, Javascript, AngularJS, Jasmine, NUnit testing

Specialization 1 : Intelligent Systems; Specialization 2 : Information Systems Grade: 1.6

Database Management Systems, File Structures, Computer Networks and Computer Organization. -Programming languages: C, C++, Java, Perl, PHP, SQL Aggregate: 82.07%