ABHISHEK BHAGWAT
adikc5@r.postjobfree.com +1-413-***-**** github.com/abhishek291994 linkedin.com/in/abhishek-bha/ EDUCATION
Northeastern University, Boston, MA Expected Dec 2020 Master’s in Information Systems
Relevant Courses: Special Topics in AI and Neural Networks, Big-Data Systems, Data Warehousing & Business Intelligence Mumbai University, Mumbai, India May 2016
Bachelor of Engineering in Computer Engineering
Certifications: Fundamentals of Deep Learning for Computer Vision NVIDIA Oracle Certified Java Professional Program (OCPJP Java SE6) PROFESSIONAL EXPERIENCE
Privafy, Boston, USA Jan 2020 – Aug 2020
Machine Learning Engineering Co-op
Behavior-based Traffic Outlier Detection System
• Built a Behavior-based Traffic Outlier Detection System to detect anomalies in network traffic to internet destinations
• Set up Spark Clusters in Google Cloud Platform(dataProc) as computing nodes for Machine Learning tasks
• Derived features by aggregating the data collected from the deployed edge devices and defined analysis dimensions
• Built unsupervised clustering model using K-means algorithm with PCA using PySpark MLlib library
• Used a rule-based algorithm to create an anomaly score based on Euclidian distance to flag suspicious behavior
• Created an ELK process to feed model output to Elastic Search using Logstash, created Kibana dashboard for investigating network anomalies
Analytics Frontend
• Developed a text-based search engine using LDA topic modeling, Elastic Search and Plotly Dash
• Built an interactive network graph driven UI to show analytics insights of time series events using Dash Cytoscape
• Set up Analytics servers to host the web app with Gunicorn and Nginx
• Web Scraped a cyber security threats website with selenium and created a threats database in Elastic Search for research Media.net Software Services, Mumbai, India June 2016 - July 2018 Research Data Analyst
• Analyzed advertising campaign data to deliver actionable insights for the Digital Marketing team
• Optimized data collection through SQL by creating views and procedures from multiple production databases
• Improved domain performances by 30% by doing Keyword analysis using Python to get highest CPC keywords
• Executed Ad-Hoc analysis, created dashboards in Tableau and reports for highest performing Domain portfolios
• Conducted AB Testing across ads and landing pages
• Deployed targeted Ad’s that improved the Yahoo Traffic Quality score to 8.0 out of 10.0 for major customer accounts TECHNICAL SKILLS
Programming Languages: Python, Java
Databases: MySQL, SQL Server, PostgreSQL
Python Libraries: NumPy, Pandas, Scikit-learn, Keras, Tensorflow, OpenCV, Plotly Dash, Flask, Selenium Machine Learning: Classification, Regression, Decision Trees, Statistical models, Data Visualization, H2o Deep Learning: Convolutional Neural Networks, RNN, LSTM, Natural Language Processing Tools and Technologies: AWS, Google Cloud Platform, Docker, Kubernetes, Elastic Search, Kibana, Tableau, Github ACADEMIC PROJECTS
Earning Calls Transcript Sentiment Analysis (GloVe, LSTM, NLP, BeautifulSoup4, AWS S3) Oct 2019
• Web Scraped financial articles using BeautifulSoup4 to build a dataset of 1600 sentiments from the latest Earnings Call Transcript (Q4 2018) for 25 companies
• Used Amazon Comprehend API to obtain the sentiment scores and stored results in AWS S3 bucket
• Implement Transfer Learning by training best model LSTM with GloVe embeddings with accuracy of 84 percent
• Created a Flask web app and containerized the application using Docker to return sentiments Image Classification on Tiny ImageNet (CNN, Keras, AutoML, Xcepetion, AWS EC2, S3) Jun 2019
• Designed a Convolutional Neural Network in Keras with Tensorflow backend for image classification on ImageNet data
• Applied Transfer Learning from the trained Tiny ImageNet model for prediction on CIFAR10 dataset in AWS
• Implemented pre-trained model Xception(78 Percent), Conducted Neural architecture search using AutoKeras Deep Learning Question Answering System (Bidirectional LSTM, Keras Tensorflow, Word Embeddings) May 2019
• Built a Deep Learning based Question Answering model on the SQUAD v2.0 dataset
• Devised a Bidirectional LSTM model with Attention, used pretrained embeddings GLOVE for question and context separately
• Engineered an outer dense layer after concatenating the separate layers to identify answers in the context Prediction on Diabetes Patient’s Hospital Readmission (SMOTES, Logistic Regression, SVM, XGBoost, Sklearn) Dec 2018
• Analyzed clinical data in that represents 10 years of clinical care at 130 US hospitals with over 50 features
• Generated Synthetic samples using SMOTES Algorithm for oversampling imbalanced class
• Predicted readmissions using Logistic Regression, Support Vector Machine and XGboost Algorithms