Nivedita Rufus

- Spearheaded the development of innovative Agentic frameworks utilizing advanced LLMs to effectively moderate sensitive and gray content on LinkedIn - Led SEO optimization initiatives for LinkedIn Articles, enhancing visibility and engagement. - Played a pivotal role in safeguarding the LinkedIn Ads space by defending against political and polarizing content during the US presidential election period. - Developed defenses against Misinformation on the LinkedIn social media platform. - Set up multiple ETL and EDA pipelines in Pyspark, and Scala from scratch for understanding the LinkedIn Ads space. - Built multi-modal(image, text, ad meta-features) AI classifiers for LinkedIn Ads to catch frauds and scams, improved from the pre-existing defense’s performance by 18% in production. - Set up modules in active learning pipelines for the AI models in production. - Set up generative-AI data augmentation pipelines for synthetic data generation to improve models's performance by 12%.

- Led the ”Software agent for estimating occupancy in medium and large buildings using RGB cameras” project - Mentored a student developer, fostering their growth and success - Achieved successful deployment and integration with RoboComp tech-stack

- Utilized machine learning, data mining, and statistical techniques to create innovative and scalable solutions for business problems. - Conducted extensive analysis to identify trends and patterns in Amazon e-commerce space to predict product popularity. - Developed Pyspark flows to analyze key metrics, such as clicks per product, click to purchase conversion ratio etc

- Developed a model for Referring Expression comprehension to identify objects of interest from English driving instructions - Referring Expression comprehension: Developed a model to identify the object of interest from a given English driving instruction. - Defined and developed baselines for Referring navigable regions, a new problem to identify navigable goal regions in images from driving instructions

- Participated in self-driving car initiative at Robotics Research Centre IIIT, Hyderabad. - Worked with Mahindra e2o electric car equipped with advanced technology. - Implemented and tested planners: RRT and Frenet frames - Hands-on experience in perception and mapping algorithms: ORB-Slam, LOAM, LeGO-LOAM. - Proficient in localization methods using GPS and IMU data.

- Project: ”Software agent for estimating occupancy in medium and large buildings using RGB cameras” - Developed a people Counter for multiple camera settings(live stream, live stream of videos from multiple views) in Python. - Laid down the future prospects for the projec

Worked on mapping, localisation(ORB-SLAM), planning (frenet frames based planner, RRT star planner).

Project: Distributed Robotic Systems Worked on robot localization and pose estimation for the multirobot formation payload transportation using IMUs.

Dealt with the internal working and emergency controls of the Pressurised Heavy water Nuclear reactor.

Dealt extensively with energy harvesting systems and also the basics of 8051 Microcontrollers.