Ritu Karivaradasamy

• Enhanced the model cost estimation tool by implementing a parallel algorithm, increasing efficiency by 50%.
• Expanded output data points, enabling teams to make more informed decisions regarding latency budget and infrastructure costs.
• Developed new user-friendly features, improving the overall usability of the tool for daily use by ML engineers.
• Authored comprehensive documentation and conducted reliability and accuracy tests, sharing results with the team to facilitate knowledge transfer and support future development.

Completed a series of projects to implement a fault-tolerant and consistent distributed system.

• Implemented chain replication with scalar clocks for consistent operation ordering and eventual consistency.
• Developed a fault-tolerant service using primary-backup replication.
• Applied Paxos protocol for replication without a single point of failure, handling network partitions.
• Built a sharded key/value store with replica groups using Paxos, emulating real-world system designs.

• Generative AI Project: Developed an ML model to automate client access management for accounting firms via QuickBooks, streamlining the recommendation process for accountants.
• Data Preparation: Crafted complex SQL queries across multiple domains and validated data to enhance model performance.
• Backend Resilience: Implemented a dead-letter message queue to prevent message buffer overflow, ensuring backend service stability.
• Frontend Quality: Implemented Playwright UI tests in ReactJS, doubling test coverage and improving the reliability of the frontend application.

• Optimized Geotracker Tool: Developed an algorithm to automatically calculate geographical boundaries with uniform sales tax rates, reducing data loading time from several days to just a few hours, significantly improving efficiency
• Enhanced Internal Tax Tool: Collaborated with the tax research team to identify customer needs, design new features, and implement improvements based on feedback. Built several APIs including event tracking and tax rate search, enabling widespread use of this data across QuickBooks teams.

• Human-Robot Interaction Research: Conducted research on robot navigation in public spaces with waiting queues. Trained a neural network to identify people in various line formations, forming the visual component of the algorithm.
• Algorithm Development: Created a heuristics-based algorithm using vector math, enabling a robot to join the end of a queue with proper distance and orientation, aligning with social cues on distance, orientation, and speed in dynamic environments.

Developed key parts of an operating system, implementing key functionalities including system calls, multi-threading and synchronization, virtual memory, and a file system.

• Female Empowerment in STEM: Mentored all-female teams in the Technovation global app development competition, empowering young girls to excel in STEM.
• Skills Development: Led workshops on iOS development using Swift and Xcode, provided guidance on debugging and feature implementation, and taught core computer science fundamentals, helping to build confidence and technical expertise among participants.

• Educational Impact on Underprivileged Kids: Contributed to StreetCode by teaching game development to kids from underrepresented communities with limited access to technology education.
• Empowering Through Technology: Guided students through the full development process—from brainstorming to testing—equipping them with valuable skills and fostering creativity in a field they might not otherwise have had the opportunity to explore.