Sowmya Sri K.

Experience

• 

  Verizon

  Dec 2018 - now

  - Working with business partners to define ML model outputs and test parameters, develop and deploy the ML models to be utilized, and evaluate the performance of those models and the experiences they are testing. - Working with the business to develop reporting that summarizes experience performance and identify opportunities for further optimization.- Working with business partners to define ML model outputs and test parameters, develop and deploy the ML models to be utilized, and evaluate the performance of those models and the experiences they are testing.- Working with domain experts, data scientists, and business partners, to identify new customer data sets that can improve adaptive model performance.- Adapting these data sets into features, and working with the Pega team to ingest new feature sets into the customer analytic record and utilizing them in adaptive model development.- Fully immersed in the business problems and solving and acting independently to move projects to completion with minimal day to day oversight.- Responsible for the entire end to end process of model development, implementation, monitoring, and maintenance of business-changing adaptive models and algorithms.-Articulating & translating business requirements to both technical and non-technical audiences.-Using Project Management tools to: establish requirements, track objectives, measure outcomes. Show less Adaptive Decision Model Analysis:-Providing analytic investigation and guidance of propositions of models across all channels and sub channels in Wire line. -Monitoring and Evaluation of Adaptive Decision Models, and report on a weekly basis the overview of Adaptive Decision Models performance on the channel level, with additional analyses as needed. - Recommend, devise and then analyze the Experimentation to test a strategy and then providing feedback which performed better or achieved the desired outcome.-Performing decomposition analysis to identify which data metric is responsible for trends or anomalies, deeper proposition analysis etc., -Automation of manual process using Python for generating the output for the experimentation, monitoring and analysis.Call Notes Automation :-The objective of the project is to reduce the call volume time when customer calls in.- Extracted negative sentiment and calling person name from customer text using Stanford NLP and NER models. -Worked with NLTK libraries to perform text classification models. Used Google News Vector models to pull the phrases of agent utterance promise made to consumer. Used cosine similarity, KMeans , LDA and experimented to increase in accuracy to predict the promise labeled to call transcripts. - Collected data, performed data cleaning, used Regular Expressions, and build machine learning models and predicted the labels. Performed classification models like Recurrent Neural Network, XGBoost. Show less

  • Sr Engr Cslt-Techinal Product Management

    Jun 2021 - now

  • Data Scientist (NLP Engineer)

    Dec 2018 - May 2021