IDENTIFICATIONbeecars.org
Brandon Carson
MS, ECEBS, Eng. Physics
DOMAINS
  • Computer Vision
  • Reinforcement Learning
  • Sensor Fusion
OCCUPATION
Artificial Intelligence Engineer
LOCATION
Seattle, WA
EMAIL
**************
LINKEDIN
/in/beecars
PROFESSIONAL SUMMARY

Applied researcher and AI engineer. Expertise deploying deep learning-based perception, estimation, and decision systems. Full ML lifecycle owner from original research to field deployment.

EXPERIENCE
Camgian · 4.5 yrsSeattle, WA
Lead Artificial Intelligence EngineerDec. 2025 – Jul. 2026
  • Led full ML lifecycle for our flagship multi-modal aerial object detection model. Achieved 0.97 mAP across aerial object classes. Integrated with robotic PTZ camera system for real-time counter-UAS operations.
  • Pioneered a multi-agent reinforcement learning method for coordinated decision-making in decentralized, partially observable environments with 100+ entities. Achieved parity with traditional optimization baselines at 1000x speedup. Integrated into field-deployable decision assistance system.
  • Delivered internal tech talks to large cross-functional audiences, covering topics such as transformers, reinforcement learning, and computer vision.
  • Mentored a team of junior engineers spanning domains of machine learning, computer vision, signal processing, and physics.
Senior Artificial Intelligence EngineerDec. 2022 – Dec. 2025
  • Research lead for a multi-modal dense image registration and fusion algorithm. Combined deep stereo disparity estimation with trifocal tensor-based correspondence transfer to achieve real-time sub-pixel alignment across optical and LWIR imaging modalities.
  • Developed an edge video surveillance system integrating object detection, activity recognition, and world-coordinate tracking. Sourced imaging sensors, constructed vision pipelines, and delivered real-time edge inference with Jetson AGX Orin using GStreamer, TensorRT, and CUDA.
  • Designed transformer-based neural architecture for probabilistic trajectory prediction models and framework for calibrating estimated covariance to real-world kinematic profiles.
  • Designed a hierarchical transformer architecture to classify time-series data with inter-series dependencies. Implemented for signal classification within a control system.
Artificial Intelligence EngineerJan. 2022 – Dec. 2022
  • Designed sensor calibration procedures for mixed-modality stereo and trinocular camera systems.
  • Designed a semi-supervised training strategy to unify disparate object detection datasets. Replaced an ensemble of detectors with single real-time detection model. Maintained mAP and decreased inference time by 70%.
  • Developed internal framework for rapid prototyping of object tracking algorithms. Integrated SotA tracker into existing vision pipeline, increased IDF1 27% and reduced per-frame runtime 10x.
University of Oklahoma · 2 yrsNorman, OK
Graduate Research AssistantAug. 2019 – Aug. 2021
  • Designed a pipeline for multi-modal 3D medical image (CT/PET) segmentation using a multi-view ensemble of UNets. Achieved 0.92 Dice score on the holdout test set, surpassing previous methods.
  • Developed an anisotropic upsampling module with learnable parameters for deep image super-resolution. Integrated into SRResNet to reconstruct low-resolution medical images, improving PSNR by 6 dB over bicubic resampling.
EDUCATION
MS, Electrical and Computer EngineeringAugust 2021

University of Oklahoma

  • Research in computer vision, deep neural networks, and image processing.
  • Carson, B.D. et al., "Approximate Vertebral Body Instance Segmentation by PET-CT Fusion for Assessment After Hematopoietic Stem Cell Transplantation", IEEE BIBE, 2023.
  • Carson, B.D., "Automatic Bone Structure Segmentation of Under-Sampled CT/FLT-PET Volumes for HSCT Patients", MS Thesis, Univ. of Oklahoma, 2021.
BS, Engineering PhysicsMay 2016

University of Oklahoma

PROJECTS

Selected Work

Write-ups are on their way.

PHOTOGRAPHY

Wildlife & Landscape