Integrations Engineer

Integrations Engineer Jemm Tec LLC Miami, FL Job Details Full-time $120,000 a year 12 hours ago Benefits Health insurance Dental insurance 401(k) Vision insurance Flexible schedule Retirement plan Qualifications AI models Jira Full-stack development Slack AI platforms (beyond public GPTs) Machine learning projects Production systems Technical writing Developing and maintaining backend systems Model training System deployment UI Defect tracking tools Figma Design (software development lifecycle) Full Job Description Date posted: May 22, 2026

Pay:

$120,000.00 per year Job description: Integrations Engineer at Jemm Tec

TO START IMMEDIATELY.

Compensation:

$120,000 per year

Location:

In-person required (Miami, FL) Reports to:

General Manager Team:

5 - 10 engineers on the Jemm Arc product

Workstation:

Jemm Tec MacBook About Jemm Tec Jemm Tec LLC (jemm.ai) builds the Jemm Arc product across hardware and software. We run on private servers, you will be given a company laptop and operate a five + layer platform stack. Our infrastructure of record lives in Jira, Confluence, and GitLab at this time. The role This is a software development role. You will build the software platform that sits between the Artificial Intelligence Engineer and the Designer at Jemm Tec, and you will be the engineer who ships the functional logic and backend systems that make their work meet inside a real product. The integrations aspect of the role: On one side, the Artificial Intelligence Engineer builds and trains the custom LLM that powers the Jemm Arc product. They own model architecture, training data, fine tunes, evaluation, and the model artifacts that come out the other end. On the other side, the Designer covers two domains. The first is the physical hardware design of the Jemm Arc itself (CAD, mechanical, enclosure, board layout thereof). The second is the web design of the user facing surfaces (mobile app, web admin console, on device UI, public product pages thereof). You sit between them and build the software platform itself. The LLM the AI Engineer trains gets injected into the platform you build. The designs the Designer produces get rendered on the user facing side of the platform you build. Your work is the functional logic, the backend systems, the API layer, the runtime that holds the model, the data plumbing, the device side software, and the companion app code thereof. Without the platform you build, the model has nowhere to live and the design has nothing to drive. Concretely, you make sure a hardware revision drawn in Fusion has the I/O the inference module needs. You make sure a mobile or web screen drawn in Figma actually calls the inference gateway with the right schema. You make sure the trained model the AI Engineer hands you runs on the device with the latency budget the product requires thereof. Without this seat, design handoffs queue up behind the AI Engineer. The AI Engineer has to context switch between training the model and wiring its outputs into firmware and product UI. The Designer's work either ships late or ships compromised because no one is the connective tissue. You are the connective tissue. You will join as one of the 5-10 engineers, report to the General Manager, and operate Claude Code (run through Cursor) as your daily AI copilot, with DeepSeek v4 available as the secondary model where it fits the task better. Owned components of the Jemm Arc product:

• JAR-7, Device Runtime and On Device AI (primary). The software shell the AI lives in on the hardware itself. On device model packaging (quantization, conversion to ONNX or Core ML or ExecuTorch thereof), inference runtime selection, device side telemetry, OTA update plumbing, and the cloud to device fallback path once a model is small enough to live on the device.
• JAR-8,

Companion Software Surfaces:

The mobile app, the web admin console, the on device touchscreen UI, and the public product pages that talk to live device data. Anything the user sees that is wired to either the hardware or the AI lives here. (Component IDs are illustrative against the existing

JAR-5 / JAR-6

map. Confirm with the Operations Manual component register before publishing thereof.) Daily routine Main project. Review and proceed with your main project operations on a daily basis. For this seat, the main project is the active Jemm Arc revision's integration milestone (the current design to device to companion delivery thereof). Jira check. Review the dashboard for tasks newly assigned by the General Manager. Slack check. Scan #engineering for the async standup thread, then scan the #feed_ channels for bot activity You will routinely take on cross team tasks of these shapes:

• "The Designer just dropped a Fusion revision for Jemm Arc Rev 5 with a new sensor on the back panel. Wire the firmware path, add the inference input channel, and update the companion app to display the new sensor field thereof."
• "The AI Engineer just finished a finetune that fits inside the 4 GB on device VRAM budget. Quantize for the device runtime, profile latency on the target SoC, and ship as an OTA release candidate."
• "The Designer's new Figma flow for first run onboarding needs three new API endpoints. Draft the OpenAPI spec, route it through #claude_proposals for review, implement the gateway, and ship the mobile and web screens."
• "Cut a release candidate of the on device runtime that gracefully falls back from local inference to cloud inference when battery thermal throttling kicks in."
• "Audit the round trip latency between the touchscreen, the on device inference runtime, and the cloud gateway. Propose two changes that take 100 ms off the worst case thereof."
• "The AI Engineer added a new tokenizer to the registry. Update the device runtime, the cloud gateway, the mobile SDK, and the web client to consume it without breaking the previous model lineage." How we work with AI tooling: Standing preapprovals-The General Manager has preapproved the following without per action approval:
• Posting summaries to #engineering from the Integrations Engineer account
• Creating subtasks under Jira epics the Integrations Engineer owns
• Updating Jira tickets assigned to the Integrations Engineer
• Drafting Confluence pages under the Integrations Engineer name (publish still requires a human click thereof)
• Posting comments on Figma files (component level review feedbackf) Everything else routes through Workflow C1.
• Slack. #engineering for technical discussion, #help_it for blockers, #incidents only for severity 1 or 2 thereof.
• Jira. Every task gets a ticket. No work happens without one, this track work done.
• Confluence. Reference docs live here once stable. Working drafts stay in Google Docs (Manual §7 document type rule thereof).
• GitLab. Branch names include the Jira ticket key (for example, JAR-712-add-sensor-channel). MR descriptions reference the ticket.
• Drive.

Working files in 02 Engineering, design assets in 03 Design, and finalized references migrate to Confluence thereof.

What we are looking for:

Production experience shipping at the boundary of hardware and software. Firmware, device drivers, or embedded Linux work alongside web or mobile work. On device AI deployment experience. This includes quantization (int8, int4, GGUF, AWQ thereof), edge inference runtimes (llama.cpp, ONNX Runtime, Core ML, ExecuTorch, TensorRT, &c.), and latency profiling on constrained hardware. Full stack Go code fluency. You will write the backend services, the device runtime, the inference gateway, the data plumbing, and the platform layer the LLM is injected into. API contract design (OpenAPI or gRPC). Comfortable drafting a spec, getting it reviewed, then implementing both sides thereof. Fluent in reading design tooling output. Figma for product, Fusion or Onshape or KiCad for hardware. You do not need to design. You need to read designs accurately and turn them into engineering work thereof. Comfortable steering an AI copilot (Claude Code inside Cursor, DeepSeek v4 inside Cursor). Strong written communication. You will translate between two roles who use different vocabularies. The bridging happens in Confluence and Jira, in writing, every day.

Nice to have:

OTA update infrastructure experience (system A/B partitions, rollback paths, signed updates thereof) BLE, MQTT, or other device to cloud protocol experience at production scale Prior work in regulated hardware contexts (FCC, CE, UL filings touched by firmware changes) Experience with a model registry (MLflow, Weights and Biases) on the consumption side. You will not train, but you will pull and ship. Fluent on macOS. We run on macOS engineering workstations and private macOS servers thereof. Familiarity with WebGL or WebGPU surfaces (Three.js, React Three Fiber). Useful for the companion app's hardware visualization views.

Benefits:

401(k) Dental insurance Flexible schedule Health insurance Retirement plan Vision insurance

Work Location:

In person

Pay:

$120,000.00 per year

Benefits:

401(k) Dental insurance Flexible schedule Health insurance Retirement plan Vision insurance

Work Location:

In person