resume_sandeepmohan

3
SANDEEP MOHAN [email protected] 1381 Pine Lake Ct (480) 370-6280 San Jose CA 95131 www.linkedin.com/in/smohan10 PROFESSIONAL SUMMARY A DSP Professional with more than 1-year experience in Speech Industry. A highly innovative professional who specializes in Signal processing and Machine learning, eager to contribute in the era of Artificial Intelligence. A 2014 Electrical and Computer Engineering Masters graduate from Arizona State University, dwells to thrive in fast-paced environment, interested in a career as Speech Processing/ Machine Learning Software Engineer. • Speech Processing • Machine Learning • C, C++ • Python • MATLAB • Probability and Statistics EDUCATION Masters (Electrical and Computer Engineering). Arizona State University (Tempe, AZ) Overall GPA: 3.57/4.0 Major: Signal Processing Graduated: 2014 Bachelors (Electronics and Communication). R.N.S. Institute of Technology (Bangalore, India) Overall Score: 83% Graduated: 2012 PROFESSIONAL SKILLS DEMONSTRATED SPEECH MORPHING SYSTEMS , INC. Nov 2014 - Present A text-to-speech startup company based in the San Francisco Bay Area DSP Software Engineer Robust Speaker Identification System: Implemented a Machine Learning algorithm to develop a Robust Speaker Identification System as a module for personalized Text to Speech service. Performed analysis of audio recording data for selecting the best features and parameters to the system. Observed the spectrogram, pitch,

Upload: sandeep-mohan

Post on 15-Apr-2017

14 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Resume_SandeepMohan

SANDEEP [email protected]

1381 Pine Lake Ct (480) 370-6280San Jose CA 95131 www.linkedin.com/in/smohan10

PROFESSIONAL SUMMARY

A DSP Professional with more than 1-year experience in Speech Industry. A highly innovative professional who specializes in Signal processing and Machine learning, eager to contribute in the era of Artificial Intelligence. A 2014 Electrical and Computer Engineering Masters graduate from Arizona State University, dwells to thrive in fast-paced environment, interested in a career as Speech Processing/ Machine Learning Software Engineer.

• Speech Processing • Machine Learning• C, C++ • Python• MATLAB • Probability and Statistics

EDUCATION

Masters (Electrical and Computer Engineering). Arizona State University (Tempe, AZ) Overall GPA: 3.57/4.0 Major: Signal Processing Graduated: 2014

Bachelors (Electronics and Communication). R.N.S. Institute of Technology (Bangalore, India) Overall Score: 83% Graduated: 2012

PROFESSIONAL SKILLS DEMONSTRATED

SPEECH MORPHING SYSTEMS, INC. Nov 2014 - PresentA text-to-speech startup company based in the San Francisco Bay Area DSP Software Engineer Robust Speaker Identification System:

Implemented a Machine Learning algorithm to develop a Robust Speaker Identification System as a module for personalized Text to Speech service.

Performed analysis of audio recording data for selecting the best features and parameters to the system. Observed the spectrogram, pitch, glottal information, formants and extracted the Mel Frequency Cepstral Coefficients along with the energies using overlapping sliding window.

Fed the features to Gaussian Mixture Model to represent a voice in a Probabilistic model. The algorithm uses the posterior probability calculation with the powerful Expectation-Maximization algorithm to converge to an optimal solution. Used unsupervised algorithm (k-means) to group similar sounding voices based on the model parameters. Later, classification was performed using i-vector based approach.

Tested the system thoroughly with the data in house. Used Scikit learn statistics to analyze the performance of the system such as Accuracy, Precision, Recall and the F-score. Prepared cases for subjective testing and provided statistics and graphs of the result.

Automatic Forced Alignment Labeling System: Tested an Automatic Labeling System which uses an HMM based Forced Alignment algorithm to extract

the labels in the audio file based on the corresponding text. The algorithm was implemented using Kaldi. Responsible for verifying the correctness of the output generated by the system. Provided a variety of

sample inputs for positive and negative test, performed memory profiling, documented the results.

Page 2: Resume_SandeepMohan

Additional information: Developed, integrated with PostgreSQL database, tested, documented and released production level

software in C++. Responsible for software maintenance and source control (Git/ Bitbucket) and bug tracking (Jira). Influenced major algorithmic & system level decision making with the CEO and VP of Engineering. Documented the system details, researched technical papers, coordinated meetings, presented and

demonstrated to various teams and investors.

RELEVANT PROJECTS

Frequency Domain Adaptive Noise Cancellation (Arizona State University): Designed an FIR adaptive filter using Least Mean Square algorithm to remove the background noise in a speech signal. Programmed functions to perform filter operations, highlighted system performance for different filter parameters. Achieved significant improvements in the Signal to Noise Ratio with speech signals with white noise and background music. (2012)

Content Based Image Retrieval (Arizona State University): Prototyped an Image Retrieval Query System based on the idea of image uniqueness and visual similarity. Used Support Vector Machines for training/classification and Histogram of Gradient for feature representation. Recorded accuracy of about 85% using the PASCALVOC 2007 dataset when tested on paintings and sketches. (2013-2014)

Motion Detection for Video Surveillance Systems (Arizona State University): Developed an algorithm to detect the presence of motion in a video. Designed Background Optimization, Alarm Trigger and Object extraction module based on the energy content in the video foreground and background. The software was written in Visual Studio C++ using OpenCV library. (2013)

CPU Architecture (Arizona State University): Realized the memory system of processor by observing the latency for each level of cache and main memory. Analyzed and evaluated the architectural insights of cache & virtual memory using CPI as a measure. Implemented multi-threaded program using pthreads and semaphore synchronization mechanism. (2013)

FPGA Research Project (WorldServe Education): Developed an algorithm for color conversion of images and optimization techniques to save memory and reduce power consumption using an FPGA Board. Performed simulations in ModelSim, synthesis, place and route, downloaded onto the Spartan 6 FPGA and performed testing. (2011-2012)

OTHER RELEVANT INFORMATION

Programming Skills, Tools & Others: C, C++, Python, MATLAB, SQL, Visual Studio C++, Linux, Numpy, Scikit-Learn, Tensor Flow, Kaldi, Boost, Statistics, Jira, Scrum, Git, PostgreSQL, Bitbucket, Verilog, FPGA, System Architecture Design, Start-up experience, Organized, Motivated, Creative & Tech enthusiast.

Awards & Certificates: Advanced Mathematics (R.N.S. Institute of Technology), Industrial Training & Mini Project in Advanced VLSI, MATLAB & FPGA Design (WorldServe Education), Data Structures and Algorithms (Coursera).