SUMMARY:

• A Deep Learning/AI Engineer with Bachelor’s degree of Applied Science in Learning technologies (Data science track) with successful background in delivering end to end solutions that address customer pain points, across domains like Recommender Systems, Anomaly detection, NLP, chatbot, Machine Vision, Collaborative Filtering, Clustering, Classification, Regression.
• Expert in statistical programming languages like R, SPSS, SAS, Dashboard Apps (R Shiny) and Python (Pandas, NumPy, Sckit - Learn, Beautiful Soup) 5for implementing machine learning algorithms in production environment. Strong Experience using OpenCV, TensorFlow, Keras, PyTorch, MXNet, Theano, Caffe and other open source frameworks.
• Actively participated in all phases of the project life cycle including data acquisition (Web Scraping), data cleaning, Data Engineering (dimensionality reduction (PCA & LDA), normalization, weight of evidence, information value), feature selection, features scaling & features engineering, Statistical modeling (decision trees, regression models, neural networks, SVM, clustering), testing and validation (ROC plot, k-fold cross validation) and data visualization.
• Experience working full data insight cycle - from discussions with business, understanding business logic and business drivers, Exploratory Data Analysis, identifying predictors, enriching data, working with missing values, exploring data dynamics, meaning or building predictive data models (if predictability can be found)
• Excellent data visualization experience either with proprietary code in R or Python, or using other visualization tools; ready for insight digestion by business and decision making to senior management.
• Extensive experience in creating visualizations and dashboards using R Shiny.
• Extensive experience using R packages like (GGPLOT2, CARET, DPLYR)
• Extensive experience in data cleaning, web scraping, fetching live streaming data, data loading & data parsing using a wide variety of Python & R packages like beautiful soup
• Hands on experience in implementing SVM, Naïve Bayes, Logistic Regression, LDA, Decision trees, Random Forests, recursive partitioning (CART), Passive Aggressive, Bagging & Boosting
• Experienced with Big Data Tools like Hadoop (HDFS),Hive
• Experience in working with Data Management and Data Governance based assignments.
• Proficient with high-level Logical Data Models, Data Mapping and Data Analysis.
• Extensive knowledge in Data Validation in Oracle and MySQL by writing SQL queries.

TECHNICAL SKILLS:

Platform: Windows, UNIX, Mac OSX, LINUX, ERP, SOA.

Databases: AWS, HADOOP (HDFS), NOSQL, SAP HANA, IBM DB2, Oracle, MS Access, MS SQL, PIG, HIVE, SPARK SQL,GCP, MICROSOFT AZURE

Language: Python, R, SAS, SQL, C, HTML, VB scripting,JAVA, Visual Studio, MATLAB

Packages: Shiny, Pandas, NUMPY, SCKIT Learn, Beautiful SOUP, GGPLOT2, CARET, MAHOUT, DPLYR, GGMCMC, ReporteS, Knitr, RJSONIO, SHinyJS, Markdown etc.

Documentation tools: LaTeX, SharePoint 2013, MS Office (Word/Excel/Power Point/ Visio),R Markdown

EXPERIENCE:

Confidential

Computer vision Engineer

Responsibilities:

• Design, develop, and implement novel computer vision algorithms for unique use cases using deep learning frameworks such as TensorFlow, keras, PyTorch, Caffe etc.
• Train neural nets to solve problems like human pose estimation, object detection, face recognition etc.
• Use tools like Tensorboard to evaluate CNN models while being trained
• Use object contours and backgrounds to generate augmented data in python
• Select and tune model hyperparameters in TensorFlow during CNN model
• Acquire and build GPUs and install necessary CUDA version and software driver
• Pre-process image dataset using tools such as OpenCV, skimage etc.
• Design and build novel/custom CNN backbones such as VGG16, Resnet, mobilenet etc.
• Evaluate accuracy and quality of the designed models as well as data sources with the help of evaluation metric like mAP, f1-score
• Design and develop production ready code in python

Confidential

Computer vision Engineer Co-op

Responsibilities:

• Design, develop, and implement novel computer vision algorithms for unique use cases using deep learning frameworks such as TensorFlow, keras, PyTorch, Caffe etc.
• Train neural nets to solve problems like human pose estimation, object detection, face recognition etc.
• Use tools like Tensorboard to evaluate CNN models while being trained
• Use object contours and backgrounds to generate augmented data in python
• Select and tune model hyperparameters in TensorFlow during CNN model
• Acquire and build GPUs and install necessary CUDA version and software driver
• Pre-process image dataset using tools such as OpenCV, skimage etc.
• Design and build novel/custom CNN backbones such as VGG16, Resnet, mobilenet etc.
• Evaluate accuracy and quality of the designed models as well as data sources with the help of evaluation metric like mAP, f1-score
• Design and develop production ready code in python

Confidential

Software Engineer Co-op

Responsibilities:

• Gathered data from Bing Streetside and openstreetmap API and trained a deep learning model on cloud GPU to localize 15 US road signs and deployed on Android and iOS platforms.
• Implemented vanishing point detection with Optical flow which improved real-time performance by 10% and dealt with jitter by implementing median filter.
• Documented software and firmware defects using bug tracking system and reported defects and procedures to ensure accurate replication which accelerated average testing turn-around time by 25%.
• Worked on lane detection pipeline for efficient Lane Departure Warning for ADAS to be used in the dashcam.

Confidential

Summer Research Fellow

Responsibilities:

• Machine learning intern at Healthcare Technology Innovation Center (HTIC), an incubation program by Confidential .
• Developed an algorithm to construct the 3D shape of the breast and screen for breast cancer on ultrasound mammograms.
• Upgraded the pre-processing of the image to remove noise by the adaptive median filtering algorithm, which improves screening accuracy by 2%. Achieved breast cancer detection by building Faster R-CNN object detection model and adjusted the hyper-parameter to increase the accuracy from 85% to 93%.
• Constructed the 3D shape of the breast using binocular stereo cameras and OpenCV to help the robotic arm to acquire the image.
• Improved the computing efficiency of the 3D reconstruction with GPU parallel computing of CUDA.

Environment: TensorFlow, Keras, Python, R, PyTorch, OpenCV, AWS