ML Research Engineer

Specter • San Francisco, CA

15d

About The Position

Specter is creating a software-defined “control plane” for the physical world. We are starting with protecting American businesses by granting them ubiquitous perception over their physical assets. To do so, we are creating a connected hardware-software ecosystem on top of multi-modal wireless mesh sensing technology. This allows us to drive down the cost and time of deploying sensors by 10x. Our platform will ultimately become the perception engine for a company’s physical footprint, enabling real-time perimeter visibility, autonomous operations management, and “digital twinning” of physical processes. Our co-founders Xerxes and Philip are passionate about empowering our partners in the fast approaching world of physical AI and robotics. We are a small, fast growing team who hail from Anduril, Tesla, Uber, and the U.S. Special Forces. Role + Responsibilities: Specter is hiring a perception AI engineer responsible for turning sensor data pipelines into actionable insights for our customers.

Requirements

• 5+ years of experience training, implementing, and deploying deep-learning based computer vision models in tasks such as object detection, semantic segmentation, object tracking, etc. (both single and multi-frame) in frameworks such as PyTorch, TensorRT, and ONNX
• Experience fine-tuning, implementing, and deploying vision-language models and large language models in frameworks such as PyTorch, TensorRT-LLM, and ONNX
• Experience optimizing model runtimes utilizing techniques such as quantization, pruning, low-rank adaptation, etc. where appropriate
• Experience building production-grade RAG pipelines, and scaling vector databases in production
• Strong experience in C++/Rust development in embedded systems and knowledge of Linux fundamentals
• Strong knowledge of CUDA fundamentals
• Experience with image/video processing, filtering, and enhancement.
• Knowledge of various video codecs desirable.
• Experience with variety of sensor types such as EO and IR cameras
• Familiarity with Rust (or ability to come up the curve quickly!)

Responsibilities

• Implementing and deploying a variety of deep-learning based vision, vision-language, and large language models to our world-class distributed perception system
• Building and scaling a production-grade data-collection, labelling, and model re-training platform
• Driving the design behind a multimodal software user interface

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed