Aug 2018 – Aug 2021 · 3 yrs · San Ramon

• Building ML solutions aimed at solving some challenging problems in Power Grids
• https://www.ge.com/digital/applications/grid-analytics
• Storm Readiness
• Improving disaster management practices in by forecasting the outages in power grid machine learning models and utilizing weather and topographical data.
• Used combination of Gradient boosting and Deep learning models. Invented product specific metrics viz Window-Accuracy. Used Keras, Tensorflow, python
• Network Connectivity
• Used anomaly detection and clustering algorithms to find connectivity mismatches and also provide alternative recommendations for mismatches. Used spark with Java to mine gigabytes of data.
• Invented meter voltage profile-based technique for detecting mismatches
• Build Supervised ML model for phase correction, solved the issue of label noise
• Product achieved 93% precision is finding mismatches and 82% precision in finding alternative recommendations. Used XGboost, python
• Case Retrieval
• Reduced the case resolution time by building answer recommender model using Bert and elastic search. Used Tensorflow, python

Dec 2017 – Jul 2018 · 7 mos · San Francisco Bay Area

• Built next-gen Data Security Tools using Natural Language Processing
• Gathered requirements, scoped out the technical stack, identified the bottlenecks and suggested solutions and created project roadmap for the final delivery
• Conducted literature review, prepared data, applied statistical and machine learning models such as Named Entity Recognition, Topic Modeling and Language models for identifying sensitive information in unstructured text
• Built a production grade, multi-class classifier using LSTMs in Tensorflow and Python, achieved 96% accuracy in identifying the sensitive information
• Built ensemble of deep learning models to detect passwords, achieved 99% accuracy
• Trained word embeddings using skip-gram model on email dataset to improve performance of predictive models

Aug 2016 – Jun 2017 · 10 mos · Bengaluru Area, India

• Led a team of 3 software engineers implementing new features in flagship mobile app in Java
• Customized Google's exoplayer library to enable digital content protection in the app
• Responsible for end-to-end product development including feature requirements, technical design, and implementation of new features. Achieved 200% increase in active user base
• Developed lead scoring model using Random Forest to prioritize sales leads which resulted in improving lead conversion ratio. Used Python

Feb 2015 – Jul 2016 · 1 yr 5 mos · Mumbai Area, India

• Designed and implemented a credit eligibility engine to offer financial products based on the profile of loan seekers. Used MVC architectures in Java viz Spring/Hibernate. Made it available to strategic alliances as a SaaS product
• With the team of 3 software developers, led the development of Rubique's flagship app from conceptualization to sign off using native Java approach
• Built state of the art functionality such as Optical Character Recognition using Google’s Tesseract library and dependency graph computation with highly maintainable architecture

Jul 2014 – Feb 2015 · 7 mos · Gurgaon, India

• Developed efficient backend API's for front end and mobile app support
• Used Java Play framework, MongoDB, Kafka, Akka, and Neo4j

Jan 2014 – Jul 2014 · 6 mos · Gurgaon, India

• • Developed analytical models for territory alignment and employee placement using SAS and VBA

Apr 2009 – Jul 2009 · 3 mos · Mumbai Area, India

• ROHYTAM: Designed the PLC based wireless controller to control Robotic trolley using a joystick, Used C, 32-bit ARM processor and Modbus networking protocol.
• Publication link: http://www.barc.gov.in/publications/eb/golden/reactor/toc/chapter16/16_6.pdf