Feb 2025 – Sep 2025 · 7 mos · Bellevue, Washington, United States · Hybrid

May 2024 – Aug 2024 · 3 mos · Bellevue, Washington, United States · On-site

Jan 2024 – May 2024 · 4 mos · Pittsburgh, Pennsylvania, United States · Remote

Sep 2023 – Mar 2024 · 6 mos

• • Spearheading collaboration between Engg and UI/UX, crafting solutions that improved experiences for all at CMU.

Jun 2023 – Aug 2023 · 2 mos · India · Remote

• Completed various product teardowns, weekly assignments, and lectures accurately, making me one of the top fellows.

Mar 2023 – Jul 2023 · 4 mos · United States · Remote

• Top 1% mentors on the platform. Mentored 100+ MS and ML aspirants.

Aug 2022 – May 2023 · 9 mos · Bengaluru, Karnataka, India

• Improving Adoption: Championed a 20% increase in DAU and 10% growth in developer base through strategic integration of TCNs, boosting OpenVino library adoption, model efficiency and accuracy by 50%.
• Cross-functional collaboration with S/W and H/W teams to gather user requirements, boosting adaptability.
• Market Research & Innovation: Pioneered the adoption of advanced neural optimization, leading to a 10% lift in predictive accuracy and empowering researchers to advance their analytical models.
• Swift Strategy Execution: Rapidly adapted business visions into product requirements, directly contributing to a 5% improvement in image classification and securing notable recognition at the Intel Innovation event.

Mar 2022 – Jul 2022 · 4 mos · East Lansing, Michigan, United States

• Worked with Prof. Sijia Liu on compression of object detection and tracking models for autonomous vehicles.
• Implemented a lightweight detection and tracking backbone for integration with state-of-the-art method QDTrack.
• Achieved a 30% increase in accuracy and a 60-fold reduction in size as compared to the QDTrack model variants.
• with Resnet-50 and Resnet-101 backbones. All models were trained on BDD100K Detection dataset.
• Developed a lightweight backbone using feature-based knowledge distillation technique with Resnet-50 as teacher network. The knowledge transfer took place on BDD100K Detection dataset.

Jun 2021 – Feb 2022 · 8 mos · Pittsburgh, Pennsylvania, United States

• Worked under the guidance of Prof. Min Xu and Dr. Sima Behpour on projects based on continual learning.
• Implemented several state-of-the-art methods such as Dynamically Expandable Networks, Orthogonal Weight
• Modification and Layerwise Optimization by Gradient Decomposition across various datasets and environments .
• Formulated and implemented a novel regularization based approach for continual learning which uses principal vectors to reconstruct the orthogonal projection matrix.
• This technique extracts only the essential gradient directions, which enhances learning as well as reduces inference.

Feb 2021 – Sep 2021 · 7 mos · Hyderabad, Telangana, India

• Product Innovation: Steered the development of a novel traffic violation detection system, integrated into police vehicles, boosting violation detection by 30% and improving public safety in a city of 50 million residents.
• Cross-Functional Leadership: Fused Engineering and UX efforts to deliver a user-centric traffic monitoring interface, facilitating its adoption by state law enforcement.
• Solution Design Precision: Engineered a unique trapezium regressor-based mechanism that accomplished 85% precision in identifying triple-riding offenses, setting a new standard in traffic regulation technology.

May 2020 – Jun 2020 · 1 mo

• Task Manager- Started and managed 10 collaborators for Google Trends task
• in order to analyse the domestic violence related keywords during COVID-19.
• Worked on regional analysis of about 200 keywords in 12 Indian languages
• which contributed in giving insights of internet usage patterns for seeking help
• and assistance during such crisis among lower tier cities in India.

Jan 2020 – Jul 2020 · 6 mos · New Delhi Area, India

• Innovation Leadership: Developed deep learning-based patented solutions for automating UAE test yards.
• Revenue Enhancement: Catalyzed a 5% increase in UAE government revenue by automating driving test
• yards, simultaneously elevating customer satisfaction.
• Data Overhead Optimization: Delivered a 95% accurate traffic-light detection system that decreased data labeling expenses by 2.5 times, reflecting a strategic approach to resource optimization.

May 2019 – Jun 2019 · 1 mo · Greater Noida

• Worked on ’Multimodal Emotion Recognition’ project under Dr.Sridhar
• Swaminathan.
• Developed LSTM based Classifier for Body Language, Facial Landmark
• Detection using Opencv and Librosa for Audio analysis of emotions and these
• were classified in 7 different categories
• Also worked on research paper on this project which was later published at IEEEBigMM Conference 2020.