MTS - Infrastructure Engineer

ironsite-ai

San Francisco (CA)

On-site

USD 175,000 - 275,000

Full time

14 days+

Application generator

A complete application in a minute — tailored resume and cover letter, ready to send.

Get past ATS filters

Benefits offered by this job

Meaningful equity at early stage

Purchase xPU/devices for development

Health, dental, vision, 401(k) with 6%

Daily catered breakfast and lunch

SF office next to Oracle Park

Job summary

Ironsite is hiring an exceptional Infrastructure Engineer to build the systems turning on-site data into the intelligence layer. You will connect rugged wearable devices to AI systems, enabling edge-to-cloud ingestion, video pipelines, and scalable research compute.

You will own the edge networks, petabyte-scale pipelines, and ML inference at the edge and in the cloud, ensuring performance, privacy, and reliability for a growing dataset and a 5M+ hours goal.

Qualifications

4+ years of experience, with 1-2 years at senior level.
Background in Computer Science, Software Engineering, Mathematics, Physics, or related field.
Experience with edge ML inference and multi-environment ML infrastructure.

Responsibilities

Build edge and device-to-cloud ingestion layer.
Develop petabyte-scale video pipelines for ingestion and processing.
Design multi-environment ML infrastructure spanning edge, on-prem, and cloud.
Ensure privacy, anonymization, and security patterns across the platform.

Skills

Infrastructure engineering

Distributed systems

Edge computing

Networking

Observability

Education

CS/SE/Math/Physics background

Tools

Edge devices

Neural accelerators

Job description

About Ironsite

Ironsite is building the intelligence layer for the physical world. We design our own wearable hardware, deploy it alongside craft workers, and transform a shift's footage into a next-morning report. Our internal team and purpose-built models label the data overnight and deliver actionable insights to superintendents by 5 AM.

We are accelerating the speed, efficiency, and predictability of construction, especially for complex, mission-critical infrastructure projects, including data centers, LNG facilities, sports stadiums, hospitals, and other large-scale developments, by training AI models on egocentric construction footage and labor productivity data. We are built with a pro-worker philosophy at our core: we believe technology should empower the workforce, not replace it. We're working to give craft workers and project leaders better visibility into what's happening on-site, while creating a system where the reality of construction and the chaos of each day is finally available to the people running the project.

Ironsite is deployed across several of the largest active construction projects in the country. To date, we've captured more than 100,000 hours of construction footage across seven states, now process thousands of hours of site activity every day, and maintain a worker opt-out rate below two percent. This is enabled by a workforce-first architecture that anonymizes devices, captures no audio, and never releases raw video.

Ironsite is backed by leading investors (8VC, South Park Commons, Saga Ventures) and prominent operators across technology and construction, including Eric Schmidt, Jeff Dean, Jeff Rothschild, Mark Leslie, Scott Wu, Eric Glyman, Karim Atiyeh, Russell Kaplan, and others, alongside over a dozen construction industry operators who have joined us as partners in building this.

Longer term, we believe Ironsite is the foundation for what construction becomes in the next decade. We think the systems we're building are the operating system for how the physical world gets built, and will unlock a fundamentally different way of respect for our workforce. One where craft workers are more valued, more visible, and better paid for the skill they bring, and where the industry finally has the intelligence layer that makes autonomous construction possible. Both futures start with the same foundation.

The Role

We're hiring an exceptional Infrastructure Engineer to build the systems that turn Ironsite's data into the intelligence layer we're building.

You'll own the backbone that connects ruggedized wearable devices in the harshest environments in American industry to the AI systems that make sense of what they capture. From the edge networks running on active jobsites, to device-to-cloud ingestion, to petabyte-scale video pipelines, to VLM inference on neural accelerators, to the compute orchestration that powers our research team. This is the infrastructure that decides how fast we can scale, how fast our models can learn, and how much of the physical world we can eventually make legible.

We're scaling toward over 5 million hours of jobsite footage in the next twelve months. The systems you build now become the substrate for everything Ironsite becomes over the next decade, from the customer-facing insights we deliver today, to the training data behind construction's first foundation models, to the eventual data layer for autonomous physical work.

What You'll Build

The edge and device-to-cloud ingestion layer. Reliable, high-throughput data transfer from wearable devices operating in low-connectivity, high-dust, high-vibration jobsites across the country. Every shift produces gigabytes of video that has to get to our infrastructure without loss, without delay, and without ever compromising worker privacy. The edge networks are part of the product, and integral to the system. You'll own them.
Petabyte-scale video pipelines. The data infrastructure that ingests, processes, stores, and serves millions of hours of first-person video for our CV and VLM teams. Designed to scale by orders of magnitude without breaking, and without leaving dead architectural weight behind.
VLM inference at the edge and in the cloud. Deploying and optimizing vision-language model inference on neural accelerator chips across edge devices, on-premises rigs, and cloud clusters. You'll design the systems that let us run the right model in the right place, with the right latency, at the right cost.
The research compute platform. Our internal research platform and multi-environment compute orchestrator is the substrate our AI team trains and evaluates on. As the operation scales toward foundation model development on 15M+ hours of egocentric construction video, this platform becomes one of the most important pieces of infrastructure at the company.
Automated reporting infrastructure. The systems that deliver AI-powered productivity insights to field leaders by 5 AM every day. That deadline is real, the reports have to be right, and the underlying pipelines have to hold up as data volume compounds.
Privacy and compliance controls. Automated face and phone blurring, worker anonymization at the individual level, and the technical guardrails that make our pro-workforce architecture real in production. These aren't features on top of the platform; they're structural properties of the platform itself.
Edge reliability at scale. IP67-rated devices operating in the harshest construction environments in the country, running continuously for 8-12 hour shifts, uploading gigabytes of video, and never failing quietly. You'll own the monitoring, observability, and resilience patterns that keep the fleet healthy at scale.

Technical Challenges You'll Solve

Video storage, distribution, and transport at YouTube-scale. As we grow from thousands of hours per day to hundreds of thousands, we'll be operating one of the more demanding first-person video infrastructures in the industry. That means solving hard problems around storage cost, retrieval latency, hot vs. cold access patterns, transport economics across the edge-to-cloud path, and the compute economics of serving video to both real-time inference pipelines and offline training runs.
Networking that has to work where nothing else does. Active construction jobsites are among the worst network environments in American industry. Miles of concrete and steel, no reliable cellular coverage, weather exposure, and crews spread across hundreds of thousands of square feet. Every one of our deployments requires field networking that has to just work, and has to keep working as the site itself changes week to week. The edge network is a product, not a utility.
Multi-environment ML infrastructure. Ironsite's compute footprint spans edge devices, on-premises rigs, cloud GPUs, and specialized neural accelerators, sometimes for a single inference request. You'll architect the systems that let us run the right workload in the right place, at the right cost, without vendor lock-in or premature abstractions.
The seams between hardware, firmware, network, and cloud. Ironsite's platform is a hardware product, a firmware product, a networking product, and a cloud product simultaneously. The most interesting infrastructure work at Ironsite lives in the seams between those layers, where a firmware quirk becomes a cloud problem, or a device fleet issue becomes a training data question, or a network hiccup on a jobsite in Texas ripples all the way through to what our researchers can train on Monday morning. You'll be the person who spans all of them, and gets to see the network and the system at their full scale.
Building infrastructure for foundation model training on data that doesn't exist yet at this scale. No one has trained a foundation model on 5M+ hours of egocentric construction video, because that data has never existed. You'll be building the training and evaluation infrastructure ahead of that curve, which means making architectural decisions today that hold up when the workload becomes real.

What We're Looking For

Deep infrastructure experience across cloud, distributed systems, and field networks. Ideally at the intersection of all three with hardware and edge devices. You've made real judgment calls on when to reach for managed services vs. build it yourself, when to optimize for cost vs. latency, and when to invest in abstraction vs. ship the boring solution.
A track record of architecting and shipping data or ML training pipelines that scaled by orders of magnitude in throughput, storage, or user count, without over-engineering early or accumulating dead architectural weight late.
Deep experience with networks in the field, in the cloud, and at their intersection. You understand what it costs to move data at scale, and you know that the network is often the most important system in the stack, and the one that quietly breaks everything if it isn't designed well.
Strong systems thinking across the full stack, from IoT and embedded, to networking, to cloud and HPC. You have the judgment to know how choices in one layer propagate up and down.
Comfort operating at the edge of what's known. You've worked on infrastructure problems where the right answer wasn't in a blog post or a textbook, and you've figured it out anyway.
Typically 4+ years of experience, with at least 1-2 spent operating at a senior level.
A background in Computer Science, Software Engineering, Mathematics, Physics, or a related technical field, or the equivalent hands-on experience.

Preferred Qualifications

Deep experience with edge ML inference and neural accelerator chips.
Background across IoT networking and traditional multi-environment HPC networking.
A history of owning the full lifecycle of systems across hardware, firmware, and software integration.
Strong instincts for observability across disciplines so you know what to measure and why.
Experience with security patterns for edge devices and video data across long lifecycles.
Genuine excitement about industries that have been underserved by software, and interest in becoming one of the engineers who builds the infrastructure the physical world runs on.

Location, Compensation, & Perks