Mar 2024 – Mar 2025 · 1 yr · Bellevue, Washington, United States · Hybrid

Jul 2023 – Mar 2024 · 8 mos · On-site

• Coordinated with configuration management team for development of a contextually aware chatbot built on real time PDF intake, which uses huggingface transformers and is deployed on streamlit. Also tried different architectures, like seq2seq, distilBERT etc.
• Reduced knowledge transfer time and onboarding time of a new employee by 12% .

May 2022 – Jul 2023 · 1 yr 2 mos

• Designed and developed a tool from scratch that facilitates code conversion, translation and migration (at scale) between different
• coding languages using AI techniques that optimized processor speed by 3.6 %.
• Improved developer efficiency by 15% by prioritizing test cases in a tool built on Bayesian Sampling thus decreasing run time of processor and increasing production confidence.

Sep 2020 – Jan 2021 · 4 mos

• Improved the performance of cloud-based systems by optimizing the allocation of resources using (DDPG) Reinforcement Learning in 6 different states of request task
• Minimised response times and relieved back-haul pressure with 40 clients and 40584 average total processed tasks in the last 10 episodes
• Implemented real-time scheduling using A3C learning & residual RNNs for decentralized learning concurrently over 4 different baselines
• Deployed experiments on the real-world dataset and improved energy consumption, response time, Service-Level-Agreement, and running cost by 14.4%, 7.74%, 31.9%, and 4.64%

Jun 2020 – Apr 2021 · 10 mos

Jan 2019 – Mar 2021 · 2 yrs 2 mos

• Devised a Multi linguistic aspect annotation using 4 psycholinguistic cues such as Binary Agreement Index, Racism score, Disposition and Dialogue acts (Allegation, Refutation, Justification) by parsing 17,000 tweets
• Conducted hate speech detection using 14 different deep learning architectures (variants of MLP, LSTM, BiLSTM, CNN) attaining the highest accuracy with CNN LSTM of 89%

May 2020 – Feb 2021 · 9 mos

• Detected Cheating in Online Chess matches using CNN - LSTM architecture with an accuracy of 89%
• Converted self-curated dataset of 90,000 games to JSON with moves represented as board states i.e (12 x 8 x 8) arrays of 1’s and 0’s
• Conceptualised nomenclature based on player reaction time per move with a special emphasis on sacrifices and mating sequences to streamline detection and improve response time by 30%

May 2020 – Jun 2020 · 1 mo · Hyderabad, Telangana, India

• Developed REST API-based scalable web app using Flask and React to display risk analysis thus reducing administrative errors
• Predicted patient’s mortality and stay using Gradient Boosting Regressor by analysing 37 parameters with a confidence of 87

Mar 2020 – Jun 2020 · 3 mos

• Analysed financial news articles with a Mining-based hierarchical classifier using 4 Performance Indicators (lagging, lagging reverse, leading, directionality) with positive and negative dictionaries for sentiment analysis
• Classified by assigning a rating of good, bad and neutral with an accuracy of 93%, 90%, 86% respectively

Sep 2019 – Jun 2020 · 9 mos · New Delhi Area, India

May 2019 – Jul 2019 · 2 mos

• Reduced the Talent Acquisition team’s burden in summarising resumes for recruitment by developing a resume management system
• composed which parses using NER and recommends Doc2Vec and Cosine Similarity;

Dec 2018 – Jan 2019 · 1 mo · New Delhi Area, India

• Worked as an web development intern at Algotraders, a DTU startup, i built a dashboard for live stock exchange and further customised it various functions for a deeper analysis of real-time data.

May 2018 – Jul 2018 · 2 mos · Oslo Area, Norway

• Predicted soil’s organic carbon content by using Random Forest, Gradient Boosting Machine and Bayesian Additive Regression Trees on 22 sampling points which explain 60-70% of variation
• Optimised spatial queries on millions of data points by using packed Hilbert R-tree.

Aug 2017 – Oct 2017 · 2 mos

• Handled social media handles(Instagram, Facebook etc) for the startup.
• Wrote Content for the posts for the same.