May 2019 – Aug 2019 · 3 mos · Greater Seattle Area

• Member of the ‘Search Speller’ team, using ML and NLP to provide world-class Speller product to billions of Facebook users.
• Updating Typeahead Speller model for Spanish users:
• Built new Phrase Ranker, Candidate Ranker and Distributed Language Models which emerged as the most important features among 670 others.
• Successfully launched the new model which beat the existing model by 10.4% Precision, 3.2% Recall, 4.9% F1-score.

Jan 2019 – Dec 2019 · 11 mos · United States

• Financial Document Analysis:
• Worked under the supervision of Prof. IV Ramakrishnan and Prof. Ritwik Banerjee on Financial Document Analysis for Broadridge company, in order to automatically extract keen information from documents which is a tedious job for humans.
• Used ML, NLP to localize and extract meeting related information from large financial documents.
• Built a successful model that is in production at Broadridge, and also presented my work to the CEO, CTO of Broadridge as well.

Jan 2019 – Dec 2019 · 11 mos · United States

• Worked as a Teaching Assistant for undergraduate students for Database Systems course and as a Grader for Theory of Computation course.
• Was responsible for interacting with students, conducting office hours, correction of papers, project evaluations and exam proctoring.

Jul 2017 – Jun 2018 · 11 mos · Hyderabad , India

• Part of the ‘Data-High Level Operating Systems’ team, majorly worked on IOT technology using C/C++ and an RTOS called ThreadX
• Multi-App and Multi-Threaded Framework: Designed and programmed an application which is multi-threaded and can be used to test all our modules parallelly. Improved the cross verifiability among different modules helping our team gain interesting insights.
• Compatibility with LLVM Compiler: Re-architectured our CRM builds and edited various .scons files to make them compatible withLLVM Compiler, in parallel with ARMCTC5.05. Widened the scope for our customers due to the open source nature of LLVM.
• Unit Testing Suite: Developed an automated, modular Unit testing framework by writing python scripts to simulate test cases from both client as well as server side. Made the system more robust as the bugs were tracked easily in a time-efficient manner.
• Framework to register AT Commands: Implemented APIs that can be used by clients to register and de-register their own AT commands, which are used to control a MODEM, with our module with a feasibility to send normal or unsolicited result code (URC) responses.

Aug 2016 – Apr 2017 · 8 mos · Hyderabad Area, India

May 2016 – Jul 2016 · 2 mos · Hyderabad Area, India

• Worked on the software development for a device, Pupil+ (Portable Pupilometer) which is a cost-effective solution to automatically detect pupillary abnormalities.
• Work included collecting data and processing the data using image processing techniques(MATLAB) followed by comparing various machine learning algorithms for automatically detecting presence of abnormal pupillary responses to light stimulation.
• GitHub Link : https://github.com/jahnavi-chowdary/PUPIL-Srujana