Srinivas Thota

Lead Data Science @Citi

Presently Working on designing and developing the Resource Cross Utilization Application for optimal utilization of resources using Machine Learning. This comprises of developing a Time Series Forecasting model for each user and forecast their productive time. Post forecasting, I have developed the Recommendation Engine for recommending the user based on skill sets from various teams. Developed various Natural Language Processing pipelines for solving the workflows with in organization. Employed some open-sourced Transfer Learning models for solving specific business use-cases. Developed OCR pipelines to capture data from documents and make it structured. Managed and developed full stack application platform which has frontend in React backend in Flask for creating multiple apps. Developed various Anomaly Detection models for detecting Anomalies in the transactions Developed String Matching algorithms for Reconciliation using Machine Learning Preprocessing techniques. Developed various NLP pipelines, which help in enhancing workflows in my organization. Developed a Language Translation pipeline using Transfer Learning. Used open sourced Facebook MBART50 model. Developed audio to text transcribe pipeline using open-sourced Facebook Wav2Vec model. Worked on Language Identification problem as part of my project. Project deals with developing a model which helps the system in identifying different languages in the text. Developed Conditional Random Field models for identifying different languages, by sourcing different feature vectors analyzed evaluation metrics. Developed Deep Neural Network Model for translating code switched language to a single language. I have used Azure extensively for Data Engineering, developed end to end data flows in finance domain. Involved in Sprint planning sessions and participated in the daily Agile SCRUM meetings. Conducted every day scrum as part of the SCRUM Master role.