OceanEye is our vessel monitoring and video recording platform, deployed across a fleet of roughly fifty vessels. It pairs an on-board recording system with cloud infrastructure on AWS that handles ingest, storage, remote access, and fleet management. It's a real production system with real constraints, and it's where you'd spend most of your time. This is a long-term engagement, not a short project — we're looking for someone who wants to own a domain and still be here in two years.
What You'd Actually Do
First 90 Days: Fleet Observability
We know more about what's happening on some vessels than others, and that asymmetry costs us. The initial focus is building the systems that show the health of every vessel — what's recording, what's reachable, what's degraded, what's about to fail — and the tooling to act on that remotely without a truck roll to a dock in another hemisphere.
Cloud Side
AWS services (Lambda, ECS, S3, Route53, IAM, CloudWatch), infrastructure as code, data pipelines, API surfaces, cost discipline on storage and egress that scales with vessel count.
Edge Side
The on-vessel stack: the recording system, IP cameras, networking over satellite, remote access tunneling, configuration management across a heterogeneous fleet where no two boats were rigged quite the same way.
You should expect to move between these in the same week — sometimes the same afternoon. The interesting problems live at the seam.
What We Need From You
Mid to Senior AWS Experience
You've built and operated production systems on AWS, not just deployed to them. You have opinions about IaC and you can defend them. You've been on the hook for an on-call rotation at some point.
Real Debugging Ability
This is the part we care most about. Given a system you've never seen, a symptom, and incomplete logs, you can form a hypothesis, isolate variables, and work down to a root cause. You know the difference between fixing a symptom and fixing a cause, and you say so out loud when you're only doing the former.
Fluency With AI-Assisted Development
We use Claude — Claude Code in particular — as a standard part of how we work. We're looking for someone who has already worked out how to get real leverage out of these tools: scoping tasks so a model can actually do them, reviewing generated code with genuine skepticism, and knowing which problems to hand off versus which to sit with yourself.
Comfort Across Linux and Windows
Our cloud side is Linux. Significant parts of the vessel side are not. You need to be functional in both without complaining much.
Networking Fundamentals
Firewalls, NAT, VPNs and zero-trust access, multicast, DNS, and an intuition for how protocols behave badly over high-latency, lossy, intermittently-available links.
Professional Written and Spoken English
A hard requirement, and a technical one. With limited overlap hours, most of the work happens in writing — issue threads, review comments, incident writeups, handoff notes. You'll also join live calls with our team and occasionally with vessel operators and partners.
We are equally uninterested in two failure modes: the engineer who won't touch AI tooling on principle, and the engineer who ships model output they don't understand. The job needs someone who can hold both.
Bonus, Not Required
Video Processing
RTSP, ONVIF, H.264/HEVC, MJPEG, ffmpeg, transcoding and bitrate tuning for constrained uplinks. Experience with NVR or VMS platforms is a real plus.
Computer Vision
Object detection and segmentation pipelines, model deployment at the edge, PyTorch.
Maritime / Marine Electronics
NMEA 0183/2000, GPS/GNSS, chartplotters, satellite connectivity (Starlink, Iridium, VSAT).
Embedded Linux / OpenWrt
Experience on small-form-factor network hardware.
Compliance-Adjacent Infrastructure
FIPS, FedRAMP, immutable evidence storage, audit trails.
How This Role Works
- Long-term independent contractor, ongoing
- Fully remote; open to candidates anywhere with 2–3 hours of daily overlap with US Eastern
- We work in Linear; expect issues, small reviewable changes, and low-ceremony process
- Contractors sign an NDA and an IP assignment agreement covering work product
Send a Note That Covers
- A production system you operated on AWS — what it did, what broke, what you changed.
- A debugging story. Something genuinely hard — the reasoning, not just the resolution.
- How you use AI tools in your work today, concretely.
- Anything from the bonus list, if it applies.
No cover letter formalities needed. Links to code, writeups, or anything you've built are welcome. Our process is short: a written exchange, then a live technical conversation where we work through a real problem together rather than quiz you on trivia.
OceanBytes is a small team. The person in this role will have direct influence over architecture decisions and will talk to the people whose boats these systems run on.
Related Reading
https://oceanbytes.com/remote-observer-services/
3 Open Source Systems Which Can Be Hosted
BoatOS
Open-source marine operating system on Raspberry Pi. View on Hackaday
Fleetbase
Open-source logistics and fleet management platform. fleetbase.io
Sea Vessel Manager
Vessel management system. seavesselmanager.com
