AI Research Engineer (Robotics)

Coram AI•Sunnyvale, CA

1d•Onsite

About The Position

At Coram AI, we’re reimagining video security for the modern world. Our cloud-native platform uses computer vision and AI to help businesses stay safe, make smarter decisions, and move faster; from real-time alerts to seamless clip sharing and multi-site visibility. You’ll be joining a small, fast-moving team that values clarity, craftsmanship, and impact. Every person here has a voice, ships meaningful work, and helps shape how AI can make the world safer and more connected. We are looking for engineers who operate at the intersection of robotics, real-time systems, and deep learning. This role focuses on deploying high-performance vision and multimodal models onto robotic platforms where latency, reliability, and hardware constraints matter.

Requirements

Strong experience building real-time robotics systems that span software and hardware
Experience deploying neural networks under strict latency constraints where milliseconds matter
Deep understanding of GPU memory management, batching strategies, and compute optimization
Strong debugging skills using profilers and low-level performance tools
Solid experience with PyTorch; experience with TensorRT and ONNX is highly desirable
Deep expertise in C++ preferred; strong Rust or Python experience also welcome
Experience building production systems that must be reliable, observable, and fault-tolerant
BS, MS, or PhD in Computer Science, Robotics, Electrical Engineering, or a related technical field
3+ years of experience building robotics, perception, or real-time systems (startup or high-performance production environments strongly preferred)
Strong programming skills in C++ (preferred) with experience in Rust or Python
Experience deploying deep learning models in production, particularly in environments with strict latency constraints
Hands-on experience with PyTorch for training, fine-tuning, and modifying deep learning models
Experience deploying models to edge devices or embedded systems where compute and memory resources are constrained
Strong debugging and profiling skills using low-level performance tools and system profilers
Experience building reliable, observable, and fault-tolerant production systems
Excellent communication skills (written and verbal) in English
Passion for building high-performance systems at the intersection of robotics, AI, and real-time infrastructure
Resilient and adaptable in challenging, fast-paced startup environments
Ability to work in an onsite environment, we move faster when we're in the same room

Nice To Haves

Experience with vLLM, SGLang, or high-performance LLM inference engines
Experience deploying multimodal models or LLMs in robotics contexts
Experience with distributed systems, structured logging, and observability at scale
Familiarity with distributed pubsub, real-time Linux, or embedded GPU platforms
Experience working with NVIDIA Jetson, CUDA kernels, or custom accelerators

Responsibilities

Deploy deep learning models on edge devices and in the cloud for real-time inference
Fine-tune models on proprietary datasets and manage dataset versioning, labeling, and evaluation
Write high-quality C++ or Rust code for deterministic, low-latency execution
Build cloud pipelines that process millions of images and video streams in near real time
Perform model surgery in PyTorch and TensorRT, including pruning, quantization, and graph optimization
Optimize GPU utilization, memory footprint, and inference throughput
Build and maintain middleware for real-time IPC between perception, planning, and control systems
Profile production systems to diagnose memory, compute, and concurrency bottlenecks
Design rigorous evaluation loops to measure model accuracy, latency, and robustness in field conditions