-
Amazon | Software Development Intern
- Engineered a centralized AWS DynamoDB data transformation mechanism, reducing write operation costs by 50%
- Designed virtual middleware to generate simplified real-time data representations for 20+ downstream teams
- Implemented fill-forwarding algorithm to standardize temporal data management across 35+ business domains
-
Association for Computing Machinery @ UCLA | Dev Team Officer
- Developed ACM's membership portal with Next.js & ExpressJS, streamlining check-ins for 1,700+ members
- Managed ACM's award-winning website on Netlify and enhancing UX, supporting UCLA's largest CS club
- Automated data integration on the Dev Team page with Javascript parser scripts, improving maintenance efficiency
-
Verifiable & Control Theoretic Robotics Lab @ UCLA | Robotics Researcher
- Implemented a LiDAR-based change detection pipeline using pose-aligned geometric differences of 3D submaps
- Built spatial octrees to query point clouds 75% faster to perform change detection in dynamic environments
- Integrated with Direct LIDAR-Inertial Odometry & Mapping algorithm, achieving long duration robot autonomy
-
Defence Science Organization (DSO) National Laboratories | Robotics Algorithm Intern
- Developed and evaluated LIDAR-Inertial Odometry (LIO) algorithms for quadrupedal robots in SLAM tasks as part of a collaboration with KAIST
- Implemented Direct LIO algorithms and degeneracy detection techniques to improve navigation in feature-sparse and motion-distorted conditions
- Tested algorithms on challenging terrains to advance legged robot autonomy and reliability
-
Laboratory for Embedded Machines & Ubiquitous Robots @ UCLA | Undergraduate Researcher
- Formulated a robust Deep Reinforcement Learning algorithm to optimize distribution of large-scale robot systems
- Employed decentralized controllers in Python to achieve Nash Equilibrium, enhancing Uber dispatching operations
- Publication - 3rd Author on “Population-aware Online Mirror Descent for Mean-Field Games by Deep Reinforcement Learning,” International Conference on Autonomous Agents and Multiagent Systems (AAMAS)
-
Defence Science Organization (DSO) National Laboratories | Robotics Algorithm Developer
- Developed Viewpoint Generation algorithm for 3D urban search, optimizing inspection for swarm robotics drones
- Built a C++ quad-tree for dynamic surface subdivision, significantly boosting inspection efficiency by 70%
- Ensured complete surface coverage by validating drone camera footprints and checking for occlusion
-
Unmanned Aerial Systems @ UCLA | UAS AUVSI Software Engineer
- Researched path-finding algorithms for fixed-wing drone to perform obstacle avoidance and autonomous mission execution
- Simulated flight behavior using ArduPilot and Gazebo to visualize navigation for Critical Design Review
- Developed Python scripts to scan airdrop boundaries for safe package delivery at 75ft in minimal time
-
Tandem Space (YC S24) | Full-Stack Developer
- Utilized React JS to streamline manual data input by 50% through intuitive entry components and listing pages
- Employed Django REST for efficient data modeling and JWT token authentication, facilitating data collection
- Developed MVP integrating front-end and back-end to enhance hybrid office space matching for 120+ companies
-
Defence Science & Technology Agency (DSTA) | Robotics Engineer
- Integrated Computer Vision and Natural Language Processing for robots in search and rescue missions
- Achieved mAP 0.5:0.95 of 83% and 98% accuracy by fine-tuning Detectron2 and HuBERT models
- Secured 5th place in national robotics competition using DJI RoboMaster S1 SDK out of 80+ teams and 3000+ students
-
Defence Science Organization (DSO) National Laboratories | Artificial Intelligence Research Intern
- Developed deep learning algorithms with over 90% accuracy in one-class classification of optical images using fast.ai
- Trained PyTorch CNN to produce descriptive features while maintaining low intra-class variance
- Optimized U-Nets for classification and devised image-cropping algorithms for preprocessing 10,000+ images
Activities & Projects
-
- Engineered an LLM-driven web-app with React JS & Material UI, reducing medication-related errors by 90%
- Leveraged Google's Gemini API, utilizing 1M tokens for precise speech-to-text and image-to-text translation
- Successfully deployed a user-friendly and real-time dashboard for medication verification, benefiting 500+ patients