Christopher Sahadeo

• Engineered a data pipeline [Python], visualizer [Rerun] and metrics to compare various off-the-shelf, SOTA feature detectors and matchers to serve as an absolute visual localization system for a GPS-denied drone prototype
• IC on a feasibility study, assessing the theoretical capabilities of various depth sensors [radar, lidar, etc] in tracking a fast, mid-sized UAV within 50cm error, to facilitate a highly timing-sensitive landing procedure
• Designed and implemented a ML pipeline [Python, Labelbox] to fine-tune and compare performance of YOLO models on domain-specific object tracking with flexibility to provision, train and infer on both local and cloud VMs
• Led a team to quickly ramp up on a production-grade codebase [React, Node.js, Prisma, PostgresSQL, FastAPI, NumPy] to implement features that control liquid handlers on a novel, machine-agnostic software platform

• Engineered a calibration station and optimizer [Jax] to calibrate multiple stereo cameras with non-overlapping fields of view, thereby increasing the accuracy of a mobile warehouse robot’s pose estimate [project info]
• Engineered many features on a calibration pipeline for a fleet of camera-based industrial robot arm work-cells, involving heavy CV work [C++ camera driver integration, calibration and maintaining a bespoke Python calibration UI]
• Led a small team in developing end-to-end, local [Golang, Bash] and GCP infrastructure design and development [project info]
• Cross-continental debugging of robot arm work cells [SSH], across the entire software stack [camera drivers on embedded Linux to cloud backend & UI]. Developed more robust logging & error handling to mitigate future failures
• Developed edge and cloud backend tools to manage configuration, control, performance, and error reporting for a fleet of industrial robot arm work-cells

• Wrapped various depth sensor SDKs [C++, Bash, Python] to operate seamlessly with a client’s existing production-grade middleware framework.
• Bootstrapped the cloud backend [AWS Amplify] and web UI [React] for an embedded, IoT small parts counting and inventory management system [project info]

Developed the simulation [ROS] and optimization [Ceres] components for a camera extrinsic calibration vSLAM system.