Dara McNally

About

Robotics engineer who built and scaled a distributed perception stack on a Unitree Go1 quadruped, coordinating 5 Jetson Nanos and a Raspberry Pi to capture, aggregate, and stream multi-camera video in real time via UDP/GStreamer and custom ROS nodes. Also implemented a YOLOv9-based detection pipeline enhanced with Grad-CAM-driven selective image enhancement (e.g., MIRNet/UFormer) to improve real-time detections and robot reactions to visual stimuli.

Experience

Education

Key Strengths

• Built distributed multi-camera sensing pipeline across 5 Jetson Nanos plus a central Raspberry Pi
• Implemented real-time video capture, streaming, and recording using UDP + GStreamer
• Developed ROS nodes to aggregate and stream all onboard camera feeds (beyond default single-stream capability)
• Designed real-time robot visual-reactivity behavior (turn-away response to detected stimulus)
• Integrated YOLOv9 with a multi-stage enhancement pipeline using Grad-CAM saliency masks and models like MIRNet/UFormer
• Performance-focused real-time optimization (sensor gating, frame-rate monitoring to manage latency)
• Self-directed ramp-up on a new robotics platform and transferred knowledge to teammates
• Emphasis on production reliability via documentation, rigorous testing, and CI/CD without breaking stable releases

Reference Highlights

• high ability in complex robotics/real-time systems
• strong interest and engagement with robotics work
• thorough testing practices
• strong debugging skills (catches uncaught errors)
• code runs reliably in field testing without errors
• extremely hard worker
• stays late to meet deliverables
• meets deadlines/deliverables on time
• asks for help appropriately
• delegates when needed
• handles ambiguity well in changing research environments
• shows initiative by proposing and designing new research projects
• effective cross-disciplinary collaborator (ML and robotics)
• Very strong in real-time systems
• Strong on the software side of complex robotics
• Produces clear, readable Python code
• Writes maintainable code (clear variable naming and comments)
• Code is easy for others to understand later
• Takes initiative through independent research
• Contributes to increasing paper publications
• Collaborates smoothly with other groups (including hardware-focused teams)
• Positive feedback on team progress speed
• Very capable at implementing real-time computer-vision systems
• Strong at refactoring and improving codebases
• Strong organizational skills in team development
• Proficient in Python (syntax, built-ins, best practices)
• Pragmatic about leveraging existing packages for speed
• Strong ownership in ambiguous environments (PhD-level independence)
• Strong work ethic and initiative
• Keeps up with fast-moving computer-vision research
• Able to build hardware-compatible, efficient models
• Capable of interfacing with various hardware/edge devices