LEAD EMBEDDED AI ENGINEER LOCATION: REMOTE DEPARTMENT: ENGINEERING · EDGE INFRASTRUCTURE REPORTS TO: CTO TYPE: FULL-TIME 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 at scale. We serve enterprise customers and we're growing fast. 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. THE ROLE We're looking for a Lead Embedded 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'll 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. WHAT YOU'LL OWN EDGE DEVICE PLATFORM * 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 CLOUD CONNECTIVITY & REMOTE MANAGEMENT * Own our AWS IoT Core integration — device provisioning, certificate management, shadow state, telemetry pipelines, and fleet-wide configuration * Design and maintain AWS Greengrass 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 CAMERA & RETAIL SYSTEMS INTEGRATION * 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 AI WORKLOAD OPTIMIZATION * Tune model inference for constrained hardware — quantization, TensorRT 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 ENGINEERING CULTURE & TOOLING * 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 WHAT WE'RE LOOKING FOR REQUIRED * 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 Greengrass — 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 Linux distributions tailored to specific hardware targets and minimal production footprints * Solid Docker experience including multi-stage builds, resource constraints, container networking, and orchestrating multiple services on resource-constrained hardware * Hands-on experience with RTSP-based camera integration and ONVIF protocol for camera discovery and management * Experience integrating with POS or other retail transaction systems at the data or protocol level * Practical experience with NVIDIA Jetson devices (Nano, Orin NX, AGX, or equivalent) and running AI inference workloads on them * Hands-on experience with Raspberry Pi Compute Module platforms (CM4 and/or CM5) in production hardware design or deployment * Proven ability to design for failure: reconnection logic, graceful degradation, remote observability, and recovery automation STRONG PLUS * Familiarity with SOC 2 environments — change management, access controls, and auditability for device fleets * Exposure to computer vision pipelines and ML model deployment beyond the hardware/runtime layer * Familiarity with hardware-aware model optimization — TensorRT, ONNX, quantization, and CPU/memory affinity configuration * Experience with retail technology ecosystems — loss prevention, CCTV, or transaction audit systems * Background in custom PCB design, carrier board selection, or hardware BOM ownership WHO YOU ARE Beyond the technical checklist, we care about how you work. You're the kind of engineer who reads error logs before asking questions. You hold a deployment in your head end-to-end — from the Python process on the device to the Greengrass component to the IoT shadow in AWS — and you notice when something doesn't add up. You don't romanticize complexity; you reduce it. And when something breaks in the field at an inconvenient time, your instinct is to get to root cause, not just restore service. We're a small, high-output team. Autonomy is real here, and so is accountability.