Mar 2022 – Oct 2025 · 3 yrs 7 mos · Tokyo, Japan · On-site

• Achievements:
• ✅ Automated Test Tracking Dashboard: Led the development of a dashboard, reducing testing time by 90% through intelligent automation.
• ✅ GenAI Application for Locator Generation: Built a GenAI-powered tool with 90% accuracy, optimizing data processing on AWS Cloud.
• ✅ Intelligent ETL System: Developed an advanced ETL pipeline, reducing project allocation time by 70%, significantly boosting operational efficiency.

Apr 2020 – Jun 2020 · 2 mos · Hyderabad, Telangana, India

• Title: Prediction of EPS for different firms
• Worked on developing a time series model for predicting a firm’s future Earning per Share(EPS) for over 5000 US-based companies by working on macroeconomic data spanning over 35 years.
• Implemented Random Forest Regressor, RNN and ARIMA on this data and the model performed better than 33.8% of Equity Analysts.
• Implemented these models on a Transition window basis which increased the percentage of Equity analysts beaten to 34.9%.
• Worked on algorithmic trading models for the Indian equity market by replicating research papers like Mohanram’s G-Score and Complicated Firms.

May 2019 – Jul 2019 · 2 mos · Kyoto, Japan

• Title: Open-Domain Question Answering
• Worked on developing a Question-Answering model using Wikipedia as the unique knowledge source.
• Used Neural Ranking model to retrieve Top 100 documents based on the query and document similarity scores from Wikipedia dump.
• Trained BERT on SQuAD (Question-Answering dataset) as the Reader Model which improved the accuracy by 4% over the DrQA model.
• Generated more Negative Samples(Questions with no answers) by using the SQuAD 2.0 dataset and then trained the Reader Model and improved the accuracy of the model by 2.9%.

May 2018 – Sep 2018 · 4 mos · Kharagpur I, India

• Title: Using social media for classifying actionable insights in a disaster scenario.
• Classifying microblogging in twitter into categories of Need Tweets and Availability Tweets during the Nepal Earthquake disaster.
• Built a TF-IDF based model using NLTK and Lucene for classification and achieved an accuracy of 86%.
• Built a model using LSTM and achieved an accuracy of 92%.

Jul 2017 – Jun 2018 · 11 mos

• Under my secretaryship B.R. Ambedkar Hall participated in various Sports event in General Championship and secured a Gold Medal in Inter Hall Football for the first time since it’s inauguration and achieved praiseworthy results in Cricket and Badminton.