Data Sales
Resume:
”
SKILLS
x Languages: Python *.*, MATLAB, Scala, R.
x Database: MySQL, PostgreSQL, SQL, MongoDB, Redis x Operating System: Linux, Windows
x Cloud: AWS, GCP, Heroku, Docker, Kubernetes
x Platforms: Google Collab, Jupyter Notebooks
x Miscellaneous: Kafka, Mongoose, GIT, Bitbucket, Kaggle, UCI Repository, ImageNet.
x Machine Learning Techniques: Supervised Learning, Unsupervised Learning, Reinforcement Learning, Linear, Logistic Regression, Decision Trees, SVM, Naïve Bayes, KNN, K-Means, Random Forest, Dimensionality Reduction
Methods, Deep Neural Networks, Cross Validation, CNN. x Image Classification Models: VGG-16, ResNet50,
Inception/ Google-Net.
x Time Series Forecasting models: Auto Regression, Moving Average, ARMA, ARIMA, SARIMA.
x Object Detection Models: R-CNN, Fast R-CNN, YOLO, SSD. x Libraries & Tools: Scikit Learn, SciPy, NumPy, Pandas, PyTorch, Scikit-Learn, TensorFlow, Matplotlib, Weka, Tableau, Keras, Caffe, Apache Spark, Theano, Jupyter Notebooks, Google Collab, Anaconda, MapReduce,
NLTK. Facebook Prophet, Spark MLlib, MXNet.
x Computer Vision, Human Pose Estimation, Time Series Forecasting/ Analysis in retail sector, Recommender Systems.
x Jetson TX-2, Raspberry-Pi
EDUCATION AND TRAINING
2020
M.S - Electrical Engineering (Machine Learning/ Data Science) San Jose State University, San Jose, CA.
2016
B.S - Electronics and Communication, India.
C O N T A C T
adhzk1@r.postjobfree.com
San Jose, CA 95113
WEBSITES, PORTFOLIOS, PROFILES
x https://www.linkedin.com/in/ustattsahi0405/
EXPERIENCE
Data Engineer, Cognizant Solutions India
09/2016 - 03/2018
x Design and build of end-to-end deep learning models in various computer vision applications as image segmentation x Performing the necessary data analysis and data pre-processing on image data to inject it to the deep learning models. x Design and build ML/DL models for time series forecasting. x Creating data pipes for deep learning models for structured and unstructured data using TensorFlow.
x Performed the necessary feature engineering on data. Data Scientist, Neuro Leap Corp San Jose, CA.
09/2018 - Present
x Used sensors and gesture recognition to observe and detect neural disabilities like Autism in young children.
x Used RESTful, GraphQL APIs for pipelining to fetch data from the cloud and feed it to our ML algorithms.
x Worked on Raspberry Pi and NVIDIA Jetson TX2 modules for machine learning applications.
x Implemented facial recognition for easy profile detection of patient. x Improved Computer Vision for recognition and tracking using machine learning techniques.
PROJECTS
Autonomous Driving 6D Image Classification
x Regression of six degrees of freedom image classification of cars with convolutional neural network, participated in live competition on Kaggle using pre-trained center net model and Pytorch library. x Classification of number of cars and identification of dimension, raw, pitch and yaw of each car is the aim of the project.
Image Classification of Kannada MNIST dataset
x Participated in Kaggle competition for classification of Kannada MNIST dataset, received a 0.97 score. x Trained model to achieve 99% test accuracy by changing no. of epochs, activation function, compiler
& hidden layers, convolution layers and kernels.
Recommender Systems
x Learned how Netflix, Spotify amazon, provide relevant search options based on the user activity. x Popularity-based recommender system.
x Content-based recommender system
x Classification-based collaborative filtering systems x Model-based collaborative filtering system
Healthcare in Machine learning: Breast Cancer Prediction x Calculated confusion matrix to measure the performance of every model and implemented x Implemented various models like Naïve Bayes, Logistic regression, Random forest, Decision tree, SVC. x Implemented data preprocessing, encoding, and scaling the data. x Also implemented keras & tensor flow library for image classification using CNN. Human Pose estimation
x Worked on multi instance pose estimation accelerated by NVIDIA TensorRT. x Low latency in running the applications was achieved. x Ran real time human pose estimation that detected features like left eye, left elbow, right ankle etc. x Worked on renet18 and densenet121 models on Jetson TX2 module with Jetpack 4.3 installed. Time Series Forecasting/ Analysis (End-To-End)
x Extracted meaningful statistics and other characteristics of the data; predicted future values based on the past values. Did trend analysis, outlier detection, stationary test, seasonality analysis. x Worked on a non-stationary dataset; retail sales and predicted how one product sales correlated with other product sales of the store.
x Visualized data using times series decomposition to see distinct components: seasonality, trend, noise. x Used ARIMA model for time series analysis and ran model diagnostics to see unusual behavior. x Used Facebook’s open source library- Prophet designed to use time series analysis and display patterns on different time scales such as yearly, weekly, daily and to see how holidays effect time series and implement custom changepoints.
SUMMARY
I am deeply interested in the domain of artificial intelligence which includes implementing machine learning models, and algorithms, statistics, probability, and programming. I have focused in the areas of retail, healthcare, computer vision, NLP and IoT. I primarily code using python language and SQL on Linux and Windows platforms. In area of machine learning, I have worked on supervised and unsupervised learning models like Linear and multiple Regression, Classification, K means clustering, Nearest neighbors and more. My expertise includes Time series Forecasting/ Analysis, computer vision and NLP. I am a very quick learner and a team player. USTATT SINGH SAHI
Contact this candidate