SATISH REDDY CHIRRA

Graduate Student in Data Science
Northeastern University, Boston, MA.

Actively seeking Full-time opportunities starting September 2019

•   View  Resume  
• Download  Resume

EDUCATION

Northeastern University, Boston, MA			    January 2017 - Present
College of Information and Computer Science 
Masters of Science in Data Science

Jawaharlal Nehru Technological University, Hyderabad, India   June 2009 - May 2013
B.Tech in Electronics and Communications

Courses

Supervised Machine Learning
Regression | Classification | Decision Trees | Ensemble Models | Neural Networks

Data Mining
Clustering | Recommendation Systems | Association rules |  Clustering | Data Wrangling

Algorithms
Divide and Conquer | Greedy Algorithms | Dynamic Programming | Sorting and Searching | Graphs

Data Visualization
D3 | Tableau | Heroku | Plotly | Matplotlib | Python | JavaScript | HTML | CSS

Data Management and Processing
R programming  | Image classification | GGPlot

Information Retrieval
Search Engine Design - Web crawling | Text acquisition and pre-processing | Indexing and Storage | Link Analysis through PageRank algorithm | User Interaction | Retrieval Models (tf-IDF, BM25, Query Likelihood) | Elastic Search | Search Result Evaluation and Ranking Techniques

Cloud Computing
Docker | Container | Kubernetes | AWS | Google Cloud Platform | JavaScript | Python

Database Management
Advanced SQL | Stored Procedures | Functions | Triggers | Database Design (Conceptual, Logical and Physical Modeling)

TECHNICAL SKILLS

Key Strengths:
Feature Engineering, Time Series Forecasting, Natural Language Processing, Data Mining, Deep Learning, Data Visualization, Statistical Modelling, ETL, Recommendation System

Languages & ML Tools:
R, Python (Scikit-learn, Numpy, Pandas, PySpark), Tensor Flow, Keras, Weka, Flask 

ML/AI Platforms:
Data Robot, H2o Driverless AI, AWS SageMaker, BigML, Azure AML

Cloud Platforms:
AWS, Google Cloud Platform, Azure, Heroku, Elasticsearch, Docker & Container Technologies

Big Data Technologies:
Spark, Hadoop

Visualization & Interpretability:
D3, Tableau, matplotlib, Plotly, ggplot, Shiny; LIME, SHAP, Global Surrogate

ETL Tools:
SAP BODS, Informatica, SAP BW, HVR

Database:
SQL (Oracle, Vector-Wise, SQL Server, Teradata), NoSQL (MongoDB), Google Big Query

Others:
Java Script, HTML, ETL Techniques, Excel, Microsoft Office and, JIRA, Service-now, Putty

PROFESSIONAL EXPERIENCE

Fidelity Investments

Data Science Coop, Boston, MA				    May 2018 - Dec. 2018

Stock Hard-to-Borrow Prediction (Multi Time-series Classification) 
Engineered time related features like Lags, Moving Averages, Difference columns and TF-IDF Vectors on Text data.
Built ML predictive models (XGBoost, LSTM) in python and used Backtesting for cross-validation and Walk forwardvalidation techniques which resulted in 1-10 days ahead prediction and better planning of inventory for brokers.
Interpreted model predictions using LIME to explain features contributing positively for the prediction to brokers.
Technologies used:  Python | XGBOOST | LSTM | Plotly | Keras | Feature Engineering | TimeSeries 

Financial Wellness Score (FWS) - eMoney 
EImplemented Financial Wellness Score for eMoney clients identifying critical categories that effect financial wellness, which helps advisors to manage clients effectively and recommend Next Best Action (NBA).
Developed interactive visualizations and animations using Plotly on various financial wellness factors.
Technologies used:  Python | Plotly | Pandas | Numpy 

Cryptocurrency Analysis
Conducted thorough exploratory data analysis to find the patterns between cryptocurrency exchanges.
Built a process flow to analyze market behavior using data from different exchanges and find the best price of bitcoin at given time.
Technologies used:  Python | Plotly | Pandas | Numpy 

Capgemini

Senior Software Engineer

Client – GE Power & Water, Bangalore, India			Feb. 2015 – Sept. 2016
Optimized ETL pipeline to improve the performance of loading financial data into data warehouse tables (Vector Wise) within 4 hours using SAP BODS from multiple sources, on which various other jobs depends.
Developed complex SQL queries to transform the data by applying the business rules before loading to DWH.
Built Tableau dashboards to visualize financial reports & managed tableau server (user management, scheduling).
Conducted UAT with the business teams to ensure the system was aligned with the vision of the business teams.
Participated in sprint planning, daily scrums, testing, retrospectives and sprint reviews.
Received ‘PAT ON BACK’ award for implementing optimization techniques in different projects.
Technologies used:  ETL | Advanced SQL | Tableau | HVR | Vector Wise | Tableau Admin | Unix | Agile | JIRA 

Software Engineer

Client – GE Capital, Bangalore, India				May 2014 – Jan. 2015
Optimized ETL Job which takes more than 90mins to load History data (134Million records) to 4mins using Parallel Partitions at session level of Informatica and also incremental (Changed data) data to 1min which loads to both Oracle and Teradata databases from a flat file.
Performed data validation by developing and executing test plans and supporting user acceptance testing.
Improved the database performance by indexing appropriate columns in various tables.
Optimized the database(oracle) by partitioning at table level for performance improvement.
Technologies used:  ETL | Oracle | SQL | Teradata | Informatica | Indexing |UAT | Agile | Unix 

ACADEMIC PROJECTS

Forecasting Retail Sales Data					Jan. 2019 – April. 2019
Built models using both the time series (ARIMA), supervised machine learning techniques (XGBoost) and Deep Learning (LSTM) leveraging different Feature Engineering techniques.
Developed visualizations for trends in sales and predictions across different locations using D3.js V4.
Designed and Developed an application ‘Sales Analyzer’ and deployed it using Heroku.
Technologies used:  D3.js V4 | Python | Tableau | Keras | JavaScript | HTML | CSS | Heroku 

Sharing Research through Data Science Environment		Jan. 2019 – April. 2019
Developed a pipeline which allows researchers/users to run the Data Science projects using Google Cloud Engine.
Automated the process of creating the virtual machine, installing the required environment and packages and running the project on cloud using Docker Images and Containers.
Collaborated in developing a front-end application where users can register their projects with the required details and keep a repository of their work along with the research papers.
Technologies used:  Google Cloud Platform (GCE) | AWS | Azure | Python | JavaScript | Docker | Containers 

Image Categorization (Stanford Dog Dataset)			Sept. 2017 – Dec. 2017
Designed and Implemented end to end Convolutional Neural Network (CNN) for 120 different dog breeds.
Fine-tuned a pre-trained model (InceptionV3) on 120 classes and achieved an accuracy of 99.16 on test data.
Technologies used:  R | EDA | Keras | CNN | Inception V3 | ggplot 

Data Mining approach to Identify Air Quality profiles		Sept. 2017 – Dec. 2017
Extracted the data from EPA Air quality data via Google cloud platform using SQL.
Implemented K-Means to cluster different pollutants across the states in USA both season and site wise. Also, used K-Median, PAM, Hierarchal, DB Scan to compare the results.
Technologies used:  Python | Google Big Query| SQL | Clustering Algorithms | Matplotlib | EDA 

Information Retrieval						May 2017 – Aug. 2017
Developed a ‘Web Scraper’ that uses BFS algorithm on the seed URLs by prioritizing the links to be crawled with the highest number of in-links and crawled above 30000 URLs in Python and indexed using Elasticsearch.
Implemented a Spam Classifier using Machine Learning Algorithm (Decision Trees) and Elasticsearch.
Technologies used:  Python | Elasticsearch | Kibana | Web Crawling | Indexing | Decision Trees | DFS | BFS 

Detection of Fake News Posts on Facebook			Jan. 2017 – April. 2017
Crawled Facebook news post comments using API and tokenized into a 300-dimensional vector space.
Trained the vectors on the corpus of comments, also used pre-trained vectors on 3 billion words from Google.
Implemented machine learning algorithms like Random Forest, Nearest Shrunken Centroid and achieved an accuracy above 90%.
Technologies used:  R | Word2Vec | Random Forest | Bayesian Inference | Google word vectors | NLP 

Sentiment Analysis on HealthCare Tweets				Jan. 2017 – April. 2017
Crawled 15000 tweets related to American Healthcare using Twitter API in R.
Processed the corpus of words using N-gram (Uni and Bi-grams) models and extracted the sentiment of the tweets that ranging from highly Positive to highly Negative and Visualized the sentiment.
Visualized data using ggplot and wordcloud.
Technologies used:  R | N-gram | Sentiment Analysis | Text Analysis | Word Cloud | ggplot 

Let's get in touch!