Github: https://github.com/ayanavasarkar LinkedIn: https://www.linkedin.com/in/ayanavasarkar/ Google Scholar: https://scholar.google.com/citations?hl=en&user=Jhs2aVYAAAAJ EDUCATION: -
- Masters in Computer
Science, University of
(Exp. Graduation May 2021)
Courses: Computer Vision,
Working under Dr. Liangliang
Cao on addressing Linguistic
Bias in VQA datasets.
Graduate Teaching Assistant
(Grader) for the course Formal
Language Theory (Spring
- Bachelors in Computer
Science, BITS, Pilani.
• Machine Learning Engineer at Imerit Technology Services. (Sept, 2018 – June, 2019)
- Used a fully-connected deep neural network pipeline to aid in the automated annotations of machine learning data of images and videos. This increased the productivity of annotators by 80%.
• Business Analyst at 64 Squares Pvt. Ltd, Pune, India. (Oct, 2017 – June, 2018)
- Built and deployed a Bi-directional Long Short-Term Memory (LSTM) to model the seasonal and weekly changes in the sale of products for an online shopping platform. This was delivered to the client using TensorFlow serving and it automated the client’s pricing system of products and directly increased their revenue.
• Visiting Research Scholar at Fulda University of Applied Sciences, Germany under the supervision of Dr. Alexander Gepperth. (Feb, 2017-Aug, 2017)
- Implemented Convolutional Self-Organizing Maps and benchmarked its performance on a GPU. This is the benchmark of all future implementations.
- Developed Gesture Recognition system for temporal video data using LSTM-RNN.
- Studied Catastrophic Forgetting using Locally Weighted Projection Regression. SKILLS: -
Languages – Python, Java.
Platforms/OS - Linux, Robot
Operating System (ROS).
Libraries - TensorFlow, Keras,
scikit-learn, SpaCy, OpenCV,
Lucene, Hadoop, Apache
• “Classifying progression of biotrophic fungus causing Rust disease in legumes”. Used pre- trained AlexNet and ResNet for classification. Bilinear pooling of the outputs from the two CNNs and adding an SD-Layer improved the performance vastly. (Paper in submission)
• Software Lead of Team IFOR. (27th June 2014 – 27th June, 2016) Work published at International Aerial Robotics Competition, 2015, Georgia, Atlanta.
Established Localization and Obstacle Avoidance using cameras and LIDAR respectively. Used Hector SLAM in ROS for localization and mapping.
Accomplished working of PTAM (Parallel Tracking and Mapping) for stabilization. ACHIEVEMENTS: -
- Reviewer for the ‘The Visual
Computer Journal’. (Impact
factor – 1.415)
- IEEEXtreme 8.0. World
Rank 444, Country Rank 4.
- Academic performance
scholarship for 2013-14.
• “An energy-based convolutional SOM model with self-adaptation capabilities”, at International Conference on Neural Networks, 2018, Greece.
• “Dynamic Hand Gesture Recognition for Mobile Systems Using Deep RNN with LSTM”, at 9th International Human Computer Interaction Conference, Paris.
• “A Collaborative Filtering based Model for Recommending Graduate Schools”, presented at 7th ICMSAO, 2017 and published in IEEE Xplore.
• “Gesture Control of Drone Using a Motion Controller” presented at The Second IEEE CIICS 2016 and published in IEEE Xplore.