Software professional with six plus years of experience in product and algorithmic development in the fields of Applied Machine Learning, Data mining/Analytics, Signal/Image/Video processing, Computer Vision, Multimedia, Touch Sensor Signal Processing and Chemometrics

Data Scientist, Human/Social Dynamics Team @ Human/Social Dynamics team @ Google • Unsupervised text clustering and topic modeling • Sentiment analysis for different brands at google • Design and implementation of auto labeling of clusters generated by topic modeling From May 2015 to Present (8 months) San Francisco Bay AreaData Science Intern @ Customer analytics Image ranking based on aesthetics, cooccurrence for improving product recommendations • Background removal using image segmentation • Automatic image annotation (tagging metadata like Color, Texture) • Semi supervised machine learning for image ranking using clustering techniques From November 2014 to April 2015 (6 months) San Francisco Bay AreaTechnical Lead @ 1) 3D location identification of finger/stylus for Capacitive Proximity Sensors in handheld devices. • Designed various features from raw self capacitance data. • Modeled a machine learning based method including Principal Component Analysis (PCA), Gaussian Mixture models (GMM), Linear Discriminant Analysis (LDA) for initial classification of the data. • Applied various regression techniques like Ordinary Least Squares (OLS), Generalized Linear Models (GLM), Errors in Variables models and Artificial Neural Networks (ANN) to estimate continuous height. • Also, designed noise simulation models to test the sensitivity of the algorithms to noisy conditions. 2) Wet touch detection and classification • Wet touch detection, localization based on horizontal/vertical profiles and watershed algorithm • Feature engineering for classifying wet touch from normal touch 3) Delivered series of presentations on Machine Learning, Patter Recognition and Regression techniques to enlighten and up-skill fellow team members in the field of machine learning 4) My responsibilities also include taking the ownership by leading the undertaken project from the front, mentoring the junior members in the team, assuring code quality etc. From April 2013 to June 2014 (1 year 3 months) Chemometrics Consultant @ Multivariate Data Analysis (Chemo-metrics) and Image Processing of Raman and Infra Red spectroscopic signals for applications ranging from biology,medicine to forensics using R, Matlab and other chemometric softwares like UnScramblerX, Origin, Wire 3.2 and Eigenvector Inc. toolboxes (PLS, MIA, EMSC). 1) Implementation of Pre-processing of Raman and FTIR signals including baseline correction (Polynomial, Asymmetric Least Squares etc), spike removal, smoothing (Savitzky-Golay, LOWESS etc), normalization (SNV, Extended Multiplicative Scatter Correction etc), Low SNR signal removal. 2) Applied various machine learning techniques for data analysis such as (a) Principal Component Analysis (PCA), Decision Trees, Linear Discriminant Analysis(LDA), Partial Least Squares (PLS), and Artificial Neural Networks (ANN) etc for classification modeling (b) Hierarchical cluster analysis(HCA), K-means, Fuzzy C-Means etc. for spatial clustering. 3) Delivered presentations on PCA, Loading Analysis and Hierarchical Clustering and their applications to Chemical Imaging 4) Worked on applications of Discrete Fourier Transform (DFT) and Linear Predictive Singular Value Decomposition (LPSVD) to Time-Resolved Raman Spectroscopy 5) Worked on building GUI for search algorithm (matching candidate spectrum with Raman spectra DataBase ) using R (rgtk2 package) From October 2012 to April 2014 (1 year 7 months) Lead Engineer @ Algorithm development for Capacitive Touch Sensing 1) Ellipse Feature Extraction for Single Ended Capacitive Touch Sensing • Extracting features like major axis length, minor axis length and angle of orientation from touch data • Used classifier based approached combined with 2d Gaussian fit features 2) Jitter Suppression for Single Ended Capacitive Touch Sensing • Devised a multi stage algorithm to suppress jitter and properly follow finger/stylus movement using time series analysis techniques including Kalman filtering, Partial Differential equations, local regression techniques and principal component analysis etc. • Implementation of the same including floating point to fixed point conversion 3) Low Noise Integration Algorithm Development for Differential Capacitive Touch Sensing • Used exponential smoothing combined with integration from both ends to suppress noise in differential domain of capacitive touch data From April 2012 to March 2013 (1 year) Senior Product Engineer @ 1) Event Detection in Broadcast Soccer Video • Locating and segmenting and tracking important foreground objects like Players, Ball and analyzing the acquired trajectories to detect and estimate parameters related to key event occurring frames • Also, devised a technique for view classification and scene change • Implementation of these algorithms as a library using OpenCV • Optimizing the code to get real time performance on 25 FPS and SD quality Video 2) Customized Video Player • Developed Multi format Video Player with growing file handling capability using ffmpeg library • Developed GUI which helps to take manual inputs and enables dynamic learning of the field and other objects in broadcast soccer video using SDL • Implemented multiple threads which interact dynamically • Capturing audio for speech recognition using ALSA library • Shared memory implementation to enable real time functioning • Implemented token passing system and Inter server communication to intelligently enable one of the operators in case of Multiple Operators (Distributed System) 3) Development of Variants of HD Audio Production System (Enhances Soccer Broadcast Audio Quality) • Got involved in requirements gathering, numerous design discussions and project planning • Developed an application which talks to Slow motion Video Servers dynamically and schedules events in case of Multi Camera environment to adapt the parameters for a given camera related to an event • Developed a GUI application which displays timeline of these events and takes manual inputs • Adapting to Live Broadcast Environment for Real-time Performance • Integration of LSA Product with BLT E’ITALIANO Video Slow motion Server • Visited IBC-2011 held at RAI, Amsterdam to understand various broadcasting systems and technology to build Outside Broadcasting (OB) Van for HD Audio Production System From October 2009 to April 2012 (2 years 7 months) Freelance Research Consultant @ 1) Defined the scope of semantics extraction in broadcast soccer video for HDAPS (Diginoc's product). 2) Developed a new background elimination technique in the context of broadcast soccer video. From April 2009 to September 2009 (6 months) Researcher @ Worked on different problems in Computer Vision • Object Detection and Classification in image and Video (funded by Samsung) • Text Detection on signboards. • Human Tracking in Surveillance Video (funded by Mind Tree) From January 2009 to September 2009 (9 months) Software Engineer @ 1) As part of applications team, worked on product named Transfort, deals with 3-D secure protocol 2) Developed scripts using AWK, SED and Customized web pages for credit unions like ANZ, Unicredit From August 2008 to December 2008 (5 months) Graduate Student @ Worked on Video to text abstraction in broadcast soccer videos by semantic analysis of interactions between key objects like ball, players From June 2007 to July 2008 (1 year 2 months)

Coursera From 2012 to 2020 Master's degree, MS in Analytics, Data Science, Big Data, 3.95/4 @ University of San Francisco From 2014 to 2015 Master's Degree, Systems Science and Automation, 6.3/8 @ Indian Institute of Science From 2006 to 2008 B.E, Electrical and Electronics Engineering, 80.1/100 @ Osmania University From 2002 to 2006 Sandeep Vanga is skilled in: Machine Learning, Signal Processing, Pattern Recognition, Image Processing, Computer Vision, Algorithms, Matlab, C, OpenCV, Python, R, C++, Linux, ARM, Embedded Software