Senior Perception Engineer, Plant Edge
Own the station pipeline: DeepStream, Holoscan, TensorRT, camera and optical synchronisation, and sub-100 ms seed, stone and check classification on Jetson-class hardware.
Careers
Work on software that moves molten glass.
Glasent is pre-launch and small. The roles below are the ones the plan needs first, each owning a loop from the plant floor to the model.
Open roles
Where we need people
All roles involve time in a glass plant. If that is unappealing, this is not the right company. These are the first roles the plan needs; openings are confirmed in the hiring conversation.
Control and Autonomy Engineer
Own the policy engine and the guarded-write path into furnace SCADA, forming and lehr controls. Experience with OT networks and IEC 62443 strongly preferred.
Simulation Engineer, Glass-line Twin
GPU-accelerated furnace CFD, glass-flow and annealing-stress simulation in Omniverse, continuously corrected against real ware data.
ML Engineer, Time Series and Prognostics
Furnace, forming and lehr drift models, melt-chemistry response and equipment prognostics from SCADA, lab and inspection history.
Robotics Engineer, Glass Handling
Isaac ROS, Isaac Manipulator and Isaac Sim for pick, transfer and stack of hot, fragile ware; sim-to-real validation before any cell goes live.
Glass Technologist
A glass technologist or process engineer who wants to encode what they know: composition windows, forming behaviour and the twin's physics.
Deployment Engineer, Plants
Take a site from rack to bounded autonomy: connectors, tag mapping, shadow mode, crew training and the baseline reconciliation.
Plant Partnerships Lead
Own design-partner relationships end to end, from the first historian extract to the signed baseline and the pilot report.
How we work
Six things about working here
Said plainly, so nobody is surprised in month two.
The plant is the office
Deployment, perception and process roles spend real time at furnaces, forming lines and lehrs. It is hot, loud and the point.
Evidence over opinion
Every claim on the site, in the product and in a pilot report traces to a record. The same standard applies inside.
Small on purpose
A deliberately small team with wide ownership. Each role owns a loop end to end.
Safety is a design requirement
Writes are bounded, logged and reversible because people stand next to this equipment. Shortcuts here are not shortcuts.
Honest about stage
Pre-launch. No customers yet, a design-partner cohort to build, and the first pilot report to earn.
Craft is the input
The technologists we work with know things no recipe wrote down. Capturing that, with their name on it, is the job.
Run timeline · scenario
What you would be building
Every Glasent run is an ordered, inspectable sequence. This is scenario_fl2_01 on FL-2 · float line · clear soda-lime: Thickness change FL-2 · 6 mm to 4 mm clear float, residual stress inside spec, zero escaped seeds. It is a worked scenario that shows the shape of a run, not a measured customer result.
scenario_fl2_01
FL-2 · float line · clear soda-lime
scenario
SUCCEEDED
-
01
ingest.order
SUCCEEDED
0.9 s
Plant Orchestrator · Pulled the 4 mm clear float spec, optical-grade tolerances and the standing energy window from plant MES; locked the target envelope for the run.
-
02
twin.simulate
SUCCEEDED
41 s
Glastwin · Simulated 36 candidate transition recipes across furnace pull, tin-bath ribbon speed and lehr curve; ranked them on seed risk, residual stress and energy per tonne.
-
03
melt.pull
SUCCEEDED
6 m 20 s
Meltrix · Stepped furnace pull toward the new ribbon mass flow while holding melt temperature and fining; chemistry stayed inside the composition window.
-
04
form.ribbon
SUCCEEDED
4 m 05 s
Formeon · Raised ribbon speed and re-angled the top rollers to thin the ribbon toward 4 mm; thickness converged inside the design tolerance.
-
05
anneal.curve
SUCCEEDED
3 m 12 s
Anneon · Re-shaped the lehr cooling curve for the thinner, faster ribbon so residual stress stays inside spec at the higher speed.
-
06
inspect.ribbon
SUCCEEDED
continuous
Seedscan · 8 camera, optical and stress stations streaming; a seed cluster flagged at the ribbon edge and attributed to the pull transient, routed to cullet.
-
07
yield.balance
SUCCEEDED
2.4 s
Plant Orchestrator · Re-sequenced cut sizes so transition ribbon routes to cullet recovery and good ribbon to the highest-value open order.
-
08
approve.human
APPROVAL
48 s
Plant Orchestrator · The second pull step exceeded the site autonomy threshold. Held for the glass technologist on shift; approved and written to the audit log.
-
09
handle.stack
SUCCEEDED
1 m 10 s
Panebot · Re-planned pick and stack paths for the thinner panes; plates flagged by Seedscan diverted to cullet, good plates stacked to rack A3.
-
10
ware.qualify
SUCCEEDED
1 m 02 s
Plant Orchestrator · Lot released with full genealogy: batch, melt, forming, lehr curve, defect map, stress map and the technologist's approval.
Process
How hiring runs
Four conversations, one piece of real work, no puzzles.
- 01Intro
Thirty minutes on what you want to own and what we need. Compensation range is shared here, not at the end.
- 02Work sample
A bounded piece of real work from the role, paid where it takes more than a few hours: a connector, a model evaluation, a tag survey, a pilot scope.
- 03Deep dive
Two hours with the people you would work with, on the sample and on the loop you would own.
- 04Plant day
Where possible, a visit to a glass plant or a working session on real plant data. Then a decision within a week.
What we offer
What we offer
Specific terms are shared in the intro conversation. The shape is this.
A loop of your own
One of the six loops, or the runtime under them, end to end, with your name on the pilot report.
The hardware to do the work
Jetson Orin on the bench, DGX or HGX time for training, OVX for the twin, real plant data under signed terms.
Compensation stated early
Salary range and equity shared in the first conversation. No negotiation theatre.
Time where the glass is
Travel to design-partner plants is part of most roles, and planned, not sprung.
Accelerated computing
Physical AI, at the furnace
A float ribbon never stops and an IS machine forms thousands of pieces an hour. Perception has to be local, deterministic and fast, so Glasent runs GPU inference at the plant edge and keeps training, simulation and optimisation on DGX, HGX and OVX.
Edge perception
4 to 24 synchronised camera, optical, thermal and stress stations per line. Sub-100 ms defect alerts on container lines and sub-500 ms fused stress and flatness decisions on flat glass are the design targets.
Model serving
Defect, stress, time-series drift and process-risk models served across edge and plant servers, with glass-knowledge and reasoning endpoints packaged as NIM services.
Training
Multimodal models fine-tuned on inspection imagery, optical and stress outputs, PLC time series, recipes, gob weight, furnace temperature, pull rate, lehr curves, energy, cullet and final grade. Planned cadence: monthly plant refreshes.
Digital twin
GPU-accelerated furnace CFD, glass-flow and annealing-stress simulation of the as-run line, with 10 to 100 candidate recipes evaluated per grade change.
Synthetic data
50,000 to 250,000 rare seed, stone, cord, inclusion, check and stress scenes per glass family, always validated against real inspection distributions before training use.
Optimisation
Pull-rate constraints, energy windows, cullet routing, forming and annealing sequence and line takt, with RAPIDS for high-volume telemetry ETL.
Autonomy
Four levels, set per agent and per tag
A plant does not go from manual to unattended in one step. Glasent makes the level explicit, auditable and reversible at any time, and the first release plan is shadow, then assist, then graduated autonomy.
| Level | What the agent does | What the person does | When |
|---|---|---|---|
| L1 · Shadow and advisory | Observes, predicts and recommends setpoints with its reasoning | Enters every change manually; a baseline is measured | Pilot weeks 1 to 3 |
| L2 · Assist | Proposes a write; it executes on approval | Approves each write in the review console | Pilot weeks 4 to 8 |
| L3 · Bounded | Writes inside tag, rate and magnitude limits on low-risk loops | Approves pull steps, grade releases and anything above threshold | Pilot week 9 onward |
| L4 · Unattended | Runs the approved envelope without prompting | Sets the envelope; reviews the shift record | Planned, after graduated autonomy proves out |
From the plant floor
The problems we hear
No customer quotes yet; we are pre-launch. These are the three buyer personas the product is built for and the pain each one describes, in their own terms.
"The furnace has run for years on the same setpoints. The process never has. We find out a melt drifted when the cullet pile grows."
Plant / operations director · ICP persona
"I can tell you why a check appeared from the lehr curve and the gob weight. I cannot be at every line, and the people who could are retiring."
Glass technologist · ICP persona
"A missed seed is a reject. A missed stress fault is a pane that shatters in the field. I need genealogy on every piece, not a spot check."
Quality / reliability engineer · ICP persona
Guardrails
An agent that can move a furnace needs a leash
Glasent writes to production equipment. Every capability is scoped, every write is policy-checked, and every action is written to an append-only audit log the plant owns.
- Bounded action space. Each agent can only write to an explicit tag allow-list, inside per-tag rate and magnitude limits.
- Policy engine before every write. Autonomy level, shift, product, interlock state and operator presence are all evaluated before a setpoint moves.
- Human-in-the-loop gates. Anything above the site threshold, pull steps, grade releases, safety-adjacent moves, waits for a named approver.
- Immutable audit log. Append-only, hash-chained, exportable, and retained on the plant's own storage.
- Hard fallback. Loss of the edge node, the network or the model returns control to the furnace, forming and lehr systems' last known-good state.
- Tenant and IP isolation. Compositions, forming recipes and defect libraries never cross a customer boundary. On-prem deployment available.
Compliance posture
| Standard | Scope | Status |
|---|---|---|
| SOC 2 Type I | Cloud control plane | RUNNING Planned in the first six months |
| SOC 2 Type II | Cloud control plane | QUEUED Planned in months six to twelve |
| IEC 62443 | Plant-edge OT security | RUNNING Design-aligned |
| ISO 9001 / IATF 16949 | Quality and genealogy records | SUCCEEDED Record formats supported |
| Container and safety-glass standards | Stress and defect conformance records | SUCCEEDED Record formats supported |
FAQ
Straight answers
The questions plant directors and glass technologists ask in the first meeting.
Yes, but only within an explicit tag allow-list with per-tag rate and magnitude limits, and only at the autonomy level your site has set. Every pilot starts in shadow mode, where Glasent predicts and recommends and a person enters everything. Writes come later, after the recommendations have earned it.
Control returns to your existing furnace, forming and lehr systems at their last known-good state. Glasent is a supervisory layer on top of the control system you already run, never a replacement for it, so an outage degrades the plant to its current way of running, not to a stop.
Shadow mode starts on the first day from existing SCADA, forming, lehr and inspection data. Defect and stress prediction improve as site history and labelled outcomes accumulate; the pilot plan sets a baseline period before any recommendation is scored.
Only if you choose cloud training. Compositions, forming recipes and defect libraries are tenant-isolated and never used to train another customer's models. On-prem training and an air-gapped plant edge are available for IP-sensitive producers.
You are, the same as with any control strategy, which is why every write is policy-checked, bounded, logged and reversible, and why anything above your risk threshold waits for a named approver. The audit log records the request, the reasoning, the limits applied and the human decision.
A 90 to 120 day line pilot in three stages: shadow mode to measure the baseline, assist mode where a technologist approves each recommendation, then bounded write-back on low-risk forming, annealing or inspection loops if the plant is satisfied with the results.
Get started
Tell us what you want to own
A role above, or one we have not listed. Write to us with the loop you would take end to end.