Computer Vision Engineer: HSV-2606

deciBel Research has an immediate opening for a Computer Vision Engineer in Dayton, OH.

Position Description:

deciBel Research is a seeking a Computer Vision Engineer to design, develop, and optimize advanced computer vision and multi‑sensor fusion algorithms supporting GEOINT‑mission workflows. You will lead the development of deep learning models, real‑time tracking systems, and GPU‑accelerated pipelines deployed in operational environments. You will shape next‑generation tools that integrate imaging physics, ML-based behavior inference, and multi‑sensor fusion. Experience developing hybrid systems that combine calibrated sensor models, deterministic algorithms, probabilistic inference, and learned components is particularly desirable. Candidates should be motivated by building systems that are scientifically explainable, empirically validated, and capable of operating robustly across both synthetic and real-world sensing environments.

Responsibilities Include:

- Develop and implement deep learning computer vision models, with a focus on sensor fusion and target tracking

- Collaborate with multidisciplinary teams to design, develop, test, and deploy technical solutions in Python or C++

- Utilize GPU programming, including CUDA or RAPIDs, to optimize the performance of computer vision applications

- Contribute to the architecture and implementation of novel single and multi-sensor detection and tracking and fusion of targets

Education Requirements:

BS in Computer Science, Electrical/Computer Engineering, AI/ML, Physics, Mathematics or another related field

Experience Requirements:

- 3+ years of experience developing computer vision algorithms for detection, tracking, or multi‑sensor fusion in remote sensing or GEOINT‑relevant environments

- 2+ years of experience applying deep learning to computer vision problems using transformer‑based or self‑supervised architectures

- Experience implementing model pipelines in Python or C++, including training, evaluation, and deployment workflows

- Knowledge of GPU‑accelerated development using CUDA, RAPIDS, or other GPU programming frameworks

- Knowledge of classical tracking or estimation methods (such as Kalman or extended filters) to support real‑time algorithm development

- Knowledge of Physics-Informed Neural Networks (PINNs), Physics-constrained loss functions

- Ability to design, test, and optimize algorithms for operational performance in constrained computing environments (i.e., multi‑GPU servers or cloud)

Special Skills Desired:

- Experience with GPU‑accelerated deep learning, including CUDA kernel development, TensorRT optimization, RAPIDS, or distributed multi‑GPU training

- Experience developing synthetic data, kinematic target models, or scenario simulation tools to support algorithm training or evaluation

- Experience with advanced estimation, tracking, and fusion techniques (such as joint multi‑sensor registration, Bayesian fusion, particle filters, deep multi‑object tracking pipelines, etc.)

- Knowledge of geospatial data formats, sensor phenomenology, or remote‑sensing exploitation workflows

- Experience with MLOps or scalable deployment systems (Docker, Kubernetes, ONNX Runtime, Triton Inference Server, or similar)

- Experience building or optimizing microservice architecture or distributed systems for real‑time data processing

- Experience integrating CV models into edge, embedded, or latency‑constrained operational environments

- Familiarity with transformer‑based vision architectures beyond DINO/CLIP/SAM (such as: ViT variants, self‑supervised multi‑modal encoders)

- Experience with cloud ML platforms (such as: AWS GovCloud, Azure ML, on-prem GPU clusters)

- Master's degree or Ph.D. in Computer Science, Electrical/Computer Engineering, AI/ML, Physics, Mathematics or another related field

Applicant selected must have an active Top Secret/SCI security clearance, with ability to take a polygraph . Must be a U.S. Citizen.