Machine Learning Engineer

ESWAR BALAJI JALAPARTHI

Boston, MA +1-857-***-**** adldqd@r.postjobfree.com Github LinkedIn EDUCATION

Northeastern University, Boston, MA Expected Graduation: May 2021 Master of Science in Data Analytics Engineering GPA:3.7/4.0 Relevant Courses: Supervised & Unsupervised Machine Learning, Deep learning, Data Warehouse and BI PGP in Data science (INSOFE), Hyderabad, India June 2017 - Jan 2018 Birla Institute of Technology (BIT), Mesra, India Aug 2013 - May 2017 Bachelor of Engineering in Mechanical Engineering

Relevant Courses: Data Structures and Algorithms, Probability and Statistics, Linear Algebra TECHNICAL SKILLS

Languages: Python, R, SQL, C++, Java

Libraries: TensorFlow, PyTorch, Keras, Scikit-learn, Numpy, Pandas, Pillow, NLTK, OpenCV Databases: MySQL, Oracle, MongoDB, S3, Hadoop, SSAS, DW Statistical Tools: Natural Language Processing, Computer Vision, Regression, Classification Tools: Docker, Tableau, GCP, Git, JIRA (Scrum Agile Methodology), Visual Studio, AWS Sagemaker PROFESSIONAL EXPERIENCE

Machine Learning Engineer at Infocrunch Analytics, Hyderabad, India Feb 2018 - Jul 2019 Social Media Analytics

• Developed an interactive social media analytics platform for generating social media reports, regarding trend and sentiments around events and products, which eliminated manual intervention and significantly reduced the turn-around time.

• Generated real time trending topics from huge amounts of social media data using LDAvis.

• Developed and deployed a Sentiment analysis model using an ensemble of pretrained models such as Deepmoji, Unsupervised-sentiment-neuron, and trained a CNN model on in-house labeled data, achieving an accuracy of 75% on three categories (positive, negative, and neutral).

• Supervised a team of 40 members to perform manual data labelling.

• Trained a transformer-based encoder decoder model to correct grammatical errors and spelling mistakes in text, leading to 3% increase in accuracy of the Sentiment analysis model. Web Services API

• Developed an OCR API, that can identify 5 Indian languages and extract text to Excel file from an image, by training an object detection model on synthetic data. Achieved a character error rate of only 5%.

• The API developed using flask was optimized to handle a huge number of requests and was hit over 30 million times and slashed the expenses to the organization by 90%. Graduate Research Assistant at Northeastern University, Boston, MA Sep 2020 - Present Forecasting COVID cases using Social Media and Geospatial data

• Utilized Web Scrapping techniques to extract ~100 GB twitter data and converted them into sparse matrices of term frequencies which reduced the memory usage by 98 %.

• Worked on spatial OLS regression and Decision Tree Regression models to predict COVID cases based on the term frequencies of N-grams and geospatial features at town level in MA. ACADEMIC PROJECTS

Intermediate task transfer

• Improved performance of several NLP tasks with less than 2000 samples, by first training BERT on an intermediate NLP task with easily available data then training on our target NLP task.

• Gained an average of ~10% improvement in performance for 10 NLP tasks with Intermediate task transfer. Gender and Age detection

• Trained multitask learning algorithm to detect age (between 5 age groups) and gender from images by finetuning Resnet101 model. Achieved 91% accuracy for gender and 74% accuracy for age detection. Datawarehouse and Business Intelligence

• Employed Talend for data profiling, building ETL pipelines to construct an enterprise Data Warehouse

(DW) subject to dimensional modelling by gathering, transforming, validating data from various data sources.

Contact this candidate