Lead Edge AI Hardware Engineer; Remote

About Panoptyc

Panoptyc is an AI-powered retail security and loss prevention platform purpose-built for the micromarket, convenience store, and enterprise retail segments. Our computer vision stack runs at the edge — directly on devices deployed in client environments — to deliver real-time shrink detection, transaction verification, and operational intelligence  serve enterprise customers and we're growing fast.

Location: Remote

Department: Engineering

• Edge Infrastructure

Reports To: CTO

Type: Full-Time

This is a high‑leverage role at the core of our physical product. The work you do here ships to real hardware in real stores, and the quality of it directly determines the reliability of the platform our customers depend on.

We're looking for a Lead Edge AI Engineer who lives at the intersection of embedded systems, cloud‑connected edge infrastructure, and computer vision. You'll own the full lifecycle of our edge device platform — from hardware selection and bring‑up through deployment pipelines, runtime orchestration, remote management, and integration with retail systems in the field.

You will work closely with our ML, backend, and product teams to ensure that inference workloads, camera feeds, and POS integrations all run reliably on constrained hardware in uncontrolled environments. This isn't a role for someone who prefers clean lab conditions — it's for someone who thrives on the complexity of the real world.

• Design, configure, and maintain edge compute solutions on Raspberry Pi CM4/CM5, NVIDIA Jetson, and similar embedded Linux platforms
• Own hardware selection and validation for new deployments, balancing compute headroom, thermal constraints, cost, and supply chain reliability
• Architect and maintain systemd service definitions for reliable, observable, auto‑recovering edge processes
• Build and manage Docker container orchestration strategies for running CV inference workloads at the edge with efficient resource utilization

• Own our AWS IoT Core integration — device provisioning, certificate management, shadow state, telemetry pipelines, and fleet‑wide configuration
• Design and maintain AWS Green grass component deployments for managing edge workloads at scale across distributed device fleets
• Build robust OTA update and rollback mechanisms that account for unreliable field connectivity

• Integrate with IP camera ecosystems using RTSP stream ingestion and ONVIF device management and discovery protocols
• Build and maintain integrations with POS systems to correlate transaction data with vision events in real time
• Ensure video pipeline reliability including reconnection logic, frame integrity checks, and latency‑aware buffering

• Tune model inference for constrained hardware — quantization, Tensor
  
  RT optimization on Jetson, ONNX runtime configuration, and CPU/GPU affinity settings
• Profile and optimize memory, thermal, and power envelopes to sustain CV workloads on edge hardware with acceptable duty cycles
• Evaluate new edge AI hardware as the landscape evolves and make informed recommendations on adoption

• Actively leverage AI coding tools and LLM‑assisted workflows as a force multiplier — this is an expectation, not a differentiator
• Document architecture, deployment runbooks, and failure modes rigorously — the team that picks up a 2am alert needs to be set up to succeed
• Collaborate across engineering, product, and installation/support teams; this role has significant cross‑functional surface area

• 5+ years of hands‑on experience with embedded Linux systems and edge hardware deployment in production environments
• Deep expertise with AWS IoT Core and AWS Green grass — device provisioning, fleet management, component deployment pipelines, and OTA updates
• Strong Python programming skills with experience writing production‑quality services and tooling (not just scripts)
• Fluency with Linux systemd — writing unit files, managing dependencies, watchdogs, journald integration, and failure recovery
• Experience with the Yocto Project for building custom embedded…