Lakshmi Manaswitha Chimakurthi

  Professional Experience

Software Engineer 				July 2019 - Present
										
Nuance Communications, mPower Clinical Analytics | Burlington, MA 
Designed and developed various advanced search features extending the mPower full text search on RAD Reports.
Designed and Implemented a highly scalable end to end system to Create and Annotate Datasets for weak labels. The assigned tags in reports are validated against a suite of Nuance AI Marketplace Algorithms and visualized the performance metrics with the Confusion Matrix.
Developed a feature for tagging the QC critical results on Radiology Reports using Nuance Clinical Language Understanding Engine and increased the detection accuracy by 15% compared to the Radiologist hardcoded detection rules.

Data Science Co-op 				May 2018 - Dec 2018
										
Channing Division of Network Medicine, Brigham and Women’s Hospital | Boston, MA 
Performed an entropy based clustering on various Lung-Tissue expression and methylation       datasets 
  Identified the cases/controls of COPD and clinical associations for each cluster
  Visualized the clustering results with ggplot in R.
Developed a docker image for cheweb (A tool for visualizing Channing’s GWAS results)
Developed a neural network with autoencoders to classify the Gold3/Gold 4 of dosage value COPD with an accuracy of 72%
Technologies used:  R | Matlab | ggplot | Python | Keras | Docker | Matplotlib 

  Projects

Sales Analyzer App	                  		      	 Jan 2019 - Apr 2019
Developed a flask application which forecasts the sales data of the store with various ML Algorithms like LSTM, Arima and XGBoost
Technologies used:  Python | Machine Learning
 https://github.com/manaswitha1001/sales-time-series-analysis

Machine Comprehension Using CommonSense Knowledge	         Jan 2018 - Apr 2018
Developed a question answering system which is trained on MCScript and DeScript Knowledge base to answer questions using related to everyday common scenarios
Technologies used:  Python | Deep Learning
 https://github.com/manaswitha1001/Machine-Comprehension-Using-Commonsense

Movie Recommender System	        			 May 2018 - July 2018
Developed a movie recommender system using collaborative filtering approach that suggests movies based on users past ratings for other movies using several ml algorithms and compared the results
Technologies used:  Python | Machine Learning
 https://github.com/manaswitha1001/predicting-movie-recommendations

Price Prediction Of Used Cars 			               April 2018 - June 2018
Developed a model using Gradient Boosting which predicts the price of the used cars using the car's attributes 
Achieved an RMSE of  .76
 deployed a working app in Heroku using Flask API
Technologies used:  Python| Scikit-Learn 
 https://github.com/manaswitha1001/Know-Your-Car-Value

Information Retrieval Systems                                   March 2018 - April 2018
Developed an Inverted Index and Search Engine over 80k document collection 
 and provided a ranked list of documents using Okapi, BM25 retrieval models for a given set of user queries.
Technologies used:  Python | Elastic Search 
 https://github.com/manaswitha1001/information-retrieval-systems

Sentiment Analysis on Customer Tweets	   April 2018
Developed a model which predicts the sentiment of the customer tweets
 regarding Airlines on Twitter
Achieved an accuracy of 78% on test data
Technologies used: Python | GLove | NLTK |Scikit-Learn
 https://github.com/manaswitha1001/Sentiment-Analysis-on-Customer-Tweets

Understand Local Business Dynamics and Neighborhood characteristics with Yelp Data			       Feb. 2018 - April 2018
Clustered the Yelp businesses data and Census data to identify how the local business dynamic patterns associate with 
 population characteristics of the neighborhood
Technologies used:  Python | Clustering | Jupyter Notebook | Census API 
 https://github.com/karantyagi/Restaurant-Recommendations-with-Yelp

Predicting Hospital Readmissions               February 2018
Developed a classifer which classifies whether a patient would be readmitted or not 
Developed using Random Forest Algorithm
Achieved an AUC of 0.68 on unseen test data 
Technologies used: Python | Scikit-Learn | Machine Learning 
 https://github.com/manaswitha1001/Predicting-Hospital-Readmissions

Boston Crime Data Analysis			          		March -2018 
Presented a poster on Boston Crime data as a part of INFORMS Data Visualization Hackathon.
Analysed the Boston crime data and visualized the top crime hotspots of the city
Analysed the top crime categories in Boston
Technologies used:  R | Plotly | Tableau 
 https://github.com/manaswitha1001/Analysis-on-Boston-Crime-Data

Prospect of Data Related Jobs in US			             	June - July 2017
Scraped the Glassdoor salaries for data related jobs across US states and stored the data in MongoDb.
Queried the database and Identified the top paying states, top hiring states and visualized the results using Plotly.
Technologies used:  R | MongoDb | Plotly 
 https://github.com/manaswitha1001/prospect-of-data-related-jobs-in-US