Python scripts for data summarization (to be used in RAF)I was in the team that was responsible for Risk Analysis and profitability prediction from early indicators. I lead the effort to predict certain benchmarks that allows our client to achieve a predefined profit in a fixed time frame. My responsibilities included data analysis, reverse engineering using regressions, guardrails creations, loss predictions using Machine Learning and writing Python scripts.Transaction Insights for AMEX- Identify merchant names from transaction description using regex functions, segmentation- grouping etc.- Vectorization (Doc2Vec etc.) and Clustering analysis. - Unigram, bigrams etc. based pattern recognition. - Identify spend and earn patterns so as to calculate a score to replace CIBIL score.

-An Artificial Intelligence based system that can access the structural health of a flapping wing employing physics based digital twin.-A computer vision based sensing approach using ML employed for real time estimation of displacement at strategically chosen locations of the wings.-A physics based digital twin of the wing created for later updating the data obtained through the monitoring of structural health and damage.-A model library with aero elastic code, based on the assumption of potential flow and linear structural behaviour, created and run for several damage cases. The information is then incorporated to the model library. -From the displacement data, damage features (natural frequency, mode shape, damping ratio etc.) extracted using auto regression.-Using supervised machine learning, damage status of the wing determined. Decision tree approach is observed to be useful in the preliminary study.

Provided detailed solutions for problems from below specified subjects of different fieldsMechanical Engineering: System Dynamics, Vibrations, Strength Of MaterialsMathematics: Numerical Methods, Complex Analysis, Sequence and Series, Probability, Linear Algebra, Integration, Vectors, Calculus, Differential EquationsComputer Science And IT: Discrete MathematicsPhysics: Modern Physics, Sound and Waves, Mechanics, Electrostatics, Optics, Gravitation

Software and ML Developer-Developed AI/ML (Machine Learning) products.-Worked on a body mapping tool, used this model to develop exercises and routines by detecting the joints of people, who performed these exercises, stitched it into routines and worked on integrating games into the website mapped with human action as opposed to using computer to play them.-Developed related logic for integration in software products.-Models which are scalable and accurate, with efficient performance on different target platforms.-Worked with ML models such as OpenPose, PoseNet and Blazepose for initial understanding of body tracking models and pose estimations.-Separate code written in JavaScript for exercise poses, animal poses, yoga poses etc, for their correct detection and also for providing further instructions, ways to correct wrong poses etc.-Tested in real time and confirmed highest accuracy along with customer satisfaction.