Prophet AI
Hardware Systems Engineer -- Vision, RFID & Edge Compute
Prophet AI, Boston, Massachusetts, us, 02298
Hardware Systems Engineer -- Vision, RFID & Edge Compute
Help us build a scalable, ruggedized sensor platform that fuses hundreds of cameras with RFID, temperature, and weight sensors to generate high‑resolution health and welfare insights in poultry production environments.
The Company: Prophet AI
In the U.S. alone, poultry mortality results in over
$4 billion in annual losses
and
8 million metric tons (MMT) of CO₂-equivalent waste , driven by undetected disease, poor welfare, and suboptimal management. Prophet AI is an early-stage AgTech company building high-resolution poultry health and welfare monitoring to eliminate this waste. We fuse large-scale video, RFID-based identity, environmental sensors (temperature, humidity), and weigh scales, to run real-time computer-vision inference on edge GPUs. We partner with breeding and animal-health companies and commercial producers to turn sensor streams into actionable, individual-level insights that improve genetics, welfare, and operational efficiency.
The Role We’re hiring a hardware systems engineer to lead the design and productization of an end‑to‑end edge platform that can reliably ingest
several hundred cameras
and
100+ additional sensors
(UHF RFID readers, temperature sensors, weight scales, etc.), while running real‑time computer vision algorithms on high-performance edge compute and syncing clean, time‑aligned data to the cloud. You’ll own architecture, vendor selection, prototyping, testing, and manufacturing hand‑off.
What You’ll Do
Own the architecture
for a multi‑sensor edge system (rack/cluster + switching + storage + power) that scales from single‑site pilots to multi‑site deployments.
Camera ingest @ scale : Design low-res camera networks (640x480p) for hundreds of streams; PoE/PoE+ budgeting; VLAN segmentation; IGMP snooping/multicast; clock sync (PTP/NTP); health monitoring and auto‑recovery.
RFID at scale : Specify and integrate ~100 UHF RFID readers/antennas; dense‑reader‑mode planning; anti‑collision tuning; shielding and interference mitigation; LLRP middleware; tag/antenna mapping and calibration.
Aux sensors : Integrate temperature, humidity, and environmental sensors and weigh scales with timestamping and calibration workflows.
Edge compute & CV pipelines : Size and tune GPU nodes; PCIe/NVLink/GPU‑direct considerations; containerized deployment; DeepStream/GStreamer pipelines; NVDEC/NVENC throughput planning.
Storage & data management : Architect local NVR‑style retention (e.g., 30–60 days) with RAID/NAS/DAS; plan write/read IOPS and sustained throughput; implement object storage gateways and offline‑first sync.
Power, thermal, reliability : Size PDUs/UPS/generators; 208/240 V distribution; rack layout, airflow, and thermal characterization; EMI/EMC and surge protection; observability (Prometheus/Grafana) and alerting.
Security & fleet ops : Network segmentation (802.1X, VLANs), certificates/PKI, secure boot, SBOM tracking; provisioning with Ansible/Terraform; OTA updates and device health.
Productization : DFM/DFT, BOM ownership and cost‑down, enclosure/industrial design, compliance (FCC/UL/CE as applicable), manufacturing partner coordination, pilot → scale playbooks.
Documentation & leadership : Write clear architecture docs, SOPs, and test plans; mentor junior engineers; collaborate with CV/ML, software, and field operations.
Minimum Qualifications
Experience building
production
multi‑sensor or video systems in industrial/retail/logistics, robotics, autonomous systems, or security/NVR at 100+ stream scale.
Deep experience with
camera systems ,
GStreamer/DeepStream ,
CUDA , and GPU resource planning (NVDEC/NVENC, memory bandwidth, PCIe lanes, NUMA).
Hands‑on with
UHF RFID
readers (e.g., Impinj/ThingMagic), antenna planning, LLRP integrations, and dense‑reader deployments.
Strong
networking
chops: L2/L3 switching, PoE budgeting, 10/25/40/100 GbE links, QoS, multicast/IGMP, PTP/NTP time sync, and site‑to‑cloud backhaul.
Proficiency in
Linux
(driver bring‑up, udev, systemd),
Docker , scripting ( Python/Bash ), and at least one systems language ( C/C++
or
Rust ).
Experience integrating
industrial sensors
and weigh scales; calibration/verification procedures.
Proven ability to take hardware from
architecture → prototype → pilot → manufacturing , with vendor selection and
BOM cost
ownership.
Nice to Have
Experience with
Triton Inference Server , Kafka/MQTT, ROS 2, or time‑series databases.
PCB/schematic literacy and lab skills (oscilloscope, spectrum analyzer); EMI/EMC troubleshooting.
Field‑hardening in harsh environments (dust, vibration, corrosive atmospheres) and IP‑rated enclosures.
Domain exposure to animal/AgTech, food production, or other regulated industrial settings.
What Success Looks Like (0–90 Days)
30 days : Detailed architecture & bill of materials for a 100+‑camera / ~100‑reader system; risk register and test plan.
60 days : Lab demo with ≥64 cameras & ≥20 RFID readers; validated ingest and CV throughput; PTP sync working across devices; storage and backhaul benchmarks.
90 days : Pilot‑ready reference rack with scripts/automation, monitoring dashboards, and install guides; factory acceptance test (FAT) checklist complete.
Pragmatic, test‑driven engineering with measurable throughput/latency/reliability targets.
Field‑first: designs must be installable, serviceable, and cost‑effective.
We value clear writing, generous collaboration, and bias‑to‑prototype.
Interview Process
1) 30‑min intro screen
2) 60‑min technical deep dive (CV/ML pipelines + networking + RFID)
3) 90‑min system design exercise
4) Panel with team
5) References
Application Send a short note, résumé/LinkedIn, and a
1‑page case study
of your largest multi‑sensor deployment (camera/RFID counts, throughput, and lessons learned) to
jean@prophetai.io
with subject
“Hardware Systems Engineer”
#J-18808-Ljbffr
The Company: Prophet AI
In the U.S. alone, poultry mortality results in over
$4 billion in annual losses
and
8 million metric tons (MMT) of CO₂-equivalent waste , driven by undetected disease, poor welfare, and suboptimal management. Prophet AI is an early-stage AgTech company building high-resolution poultry health and welfare monitoring to eliminate this waste. We fuse large-scale video, RFID-based identity, environmental sensors (temperature, humidity), and weigh scales, to run real-time computer-vision inference on edge GPUs. We partner with breeding and animal-health companies and commercial producers to turn sensor streams into actionable, individual-level insights that improve genetics, welfare, and operational efficiency.
The Role We’re hiring a hardware systems engineer to lead the design and productization of an end‑to‑end edge platform that can reliably ingest
several hundred cameras
and
100+ additional sensors
(UHF RFID readers, temperature sensors, weight scales, etc.), while running real‑time computer vision algorithms on high-performance edge compute and syncing clean, time‑aligned data to the cloud. You’ll own architecture, vendor selection, prototyping, testing, and manufacturing hand‑off.
What You’ll Do
Own the architecture
for a multi‑sensor edge system (rack/cluster + switching + storage + power) that scales from single‑site pilots to multi‑site deployments.
Camera ingest @ scale : Design low-res camera networks (640x480p) for hundreds of streams; PoE/PoE+ budgeting; VLAN segmentation; IGMP snooping/multicast; clock sync (PTP/NTP); health monitoring and auto‑recovery.
RFID at scale : Specify and integrate ~100 UHF RFID readers/antennas; dense‑reader‑mode planning; anti‑collision tuning; shielding and interference mitigation; LLRP middleware; tag/antenna mapping and calibration.
Aux sensors : Integrate temperature, humidity, and environmental sensors and weigh scales with timestamping and calibration workflows.
Edge compute & CV pipelines : Size and tune GPU nodes; PCIe/NVLink/GPU‑direct considerations; containerized deployment; DeepStream/GStreamer pipelines; NVDEC/NVENC throughput planning.
Storage & data management : Architect local NVR‑style retention (e.g., 30–60 days) with RAID/NAS/DAS; plan write/read IOPS and sustained throughput; implement object storage gateways and offline‑first sync.
Power, thermal, reliability : Size PDUs/UPS/generators; 208/240 V distribution; rack layout, airflow, and thermal characterization; EMI/EMC and surge protection; observability (Prometheus/Grafana) and alerting.
Security & fleet ops : Network segmentation (802.1X, VLANs), certificates/PKI, secure boot, SBOM tracking; provisioning with Ansible/Terraform; OTA updates and device health.
Productization : DFM/DFT, BOM ownership and cost‑down, enclosure/industrial design, compliance (FCC/UL/CE as applicable), manufacturing partner coordination, pilot → scale playbooks.
Documentation & leadership : Write clear architecture docs, SOPs, and test plans; mentor junior engineers; collaborate with CV/ML, software, and field operations.
Minimum Qualifications
Experience building
production
multi‑sensor or video systems in industrial/retail/logistics, robotics, autonomous systems, or security/NVR at 100+ stream scale.
Deep experience with
camera systems ,
GStreamer/DeepStream ,
CUDA , and GPU resource planning (NVDEC/NVENC, memory bandwidth, PCIe lanes, NUMA).
Hands‑on with
UHF RFID
readers (e.g., Impinj/ThingMagic), antenna planning, LLRP integrations, and dense‑reader deployments.
Strong
networking
chops: L2/L3 switching, PoE budgeting, 10/25/40/100 GbE links, QoS, multicast/IGMP, PTP/NTP time sync, and site‑to‑cloud backhaul.
Proficiency in
Linux
(driver bring‑up, udev, systemd),
Docker , scripting ( Python/Bash ), and at least one systems language ( C/C++
or
Rust ).
Experience integrating
industrial sensors
and weigh scales; calibration/verification procedures.
Proven ability to take hardware from
architecture → prototype → pilot → manufacturing , with vendor selection and
BOM cost
ownership.
Nice to Have
Experience with
Triton Inference Server , Kafka/MQTT, ROS 2, or time‑series databases.
PCB/schematic literacy and lab skills (oscilloscope, spectrum analyzer); EMI/EMC troubleshooting.
Field‑hardening in harsh environments (dust, vibration, corrosive atmospheres) and IP‑rated enclosures.
Domain exposure to animal/AgTech, food production, or other regulated industrial settings.
What Success Looks Like (0–90 Days)
30 days : Detailed architecture & bill of materials for a 100+‑camera / ~100‑reader system; risk register and test plan.
60 days : Lab demo with ≥64 cameras & ≥20 RFID readers; validated ingest and CV throughput; PTP sync working across devices; storage and backhaul benchmarks.
90 days : Pilot‑ready reference rack with scripts/automation, monitoring dashboards, and install guides; factory acceptance test (FAT) checklist complete.
Pragmatic, test‑driven engineering with measurable throughput/latency/reliability targets.
Field‑first: designs must be installable, serviceable, and cost‑effective.
We value clear writing, generous collaboration, and bias‑to‑prototype.
Interview Process
1) 30‑min intro screen
2) 60‑min technical deep dive (CV/ML pipelines + networking + RFID)
3) 90‑min system design exercise
4) Panel with team
5) References
Application Send a short note, résumé/LinkedIn, and a
1‑page case study
of your largest multi‑sensor deployment (camera/RFID counts, throughput, and lessons learned) to
jean@prophetai.io
with subject
“Hardware Systems Engineer”
#J-18808-Ljbffr