Akshay Khare Email & Phone Number

Bengaluru, KA, IN

Solved challenging use cases in the Document AI domain such as key-value pair information extraction and table detection

Hyderabad, Telangana, India

Worked on interesting Computer Vision problems in the sports domain. Developed a wide variety of state-of-the-art deep learning models for use cases ranging from object detection and tracking to event detection and pose estimation.

Chennai, Tamil Nadu, India

• Advisor: Prof. Harish G Ramaswamy, Department of Computer Science, IIT Madras
• Developed a novel domain adaptation approach that involves training multiple diverse classifiers to improve generalization on test data that exhibits a distributional shift from training data
• Formulated new loss functions that promote model diversity. Achieved an accuracy uplift of 25% over training a single classifier, for certain types of biases in the data

Manchester, United Kingdom

• Advisor: Prof. Philip Scarf
• Developed novel statistical models to predict Australian Football League scores, estimated model parameters by maximising discounted log likelihood loss
• Devised a game simulation model to visualize time evolution of win probability which serves to qualitatively assess outcome uncertainty
• Published a journal paper that proposes a broad framework to analyse outcome uncertainty in sports

Pune, Maharashtra, India

• Developed a computer vision system for real-time detection and tracking of vehicles in CCTV footage.
• Devised a novel metric for image similarity using a combination of multiple feature extractors.

Pune, Maharashtra, India

• Developed an email classification system to automatically segregate customer support emails, thus facilitating assistance to customers at increased efficiency
• Built a bag-of-words model utilizing TF-IDF scheme to extract words with high discriminatory power
• Achieved content-based email classification with F-1 score of 92.9%

Cochin Area, India