Who are we?
We are a team of driven individuals who strongly believe in the positive impact gaming and AI will have on the learning world. Today, we're helping enterprise businesses in high-risk industries move to immersive, gamified safety training that far outperforms its traditional counterparts, so every frontline worker goes home safely. If you strive to work in a creative and high-output environment, read on.
Who are we looking for?
An AI Video Pipeline Engineer who is a technical artist first and a pipeline builder second. You have the eye of an artist and the instincts of a systems thinker, and you have probably made the jump from Unreal or Unity into open and closed weight models , ControlNets, higgsfield or similar and ComfyUI. You will build and own the AI image and video pipeline the whole studio uses, turning R&D into a working system that makes our output faster, more consistent and higher quality.
You'll report to Ryan, who leads our AI research, and take what he's exploring and make it something every artist reaches for by default. We haven't locked our direction between open-weight and closed-weight tools, and we don't want to. We want someone with strong, evidence-backed opinions who can debate it with us and help make the call.
You will be responsible for:
Building, owning and maintaining the AI image and video pipeline the whole team uses, across our LFI (Learning From Incidents) reconstructions, 2D image work and video output
Solving consistency at scale: the same character, the same site and the same style from shot to shot and module to module, using trained LoRAs, ControlNets, IP-Adapters and reference workflows
Training and maintaining models on our own characters, worlds and styles, so our cast and environments hold up in every frame
Turning our research into production: taking what's being explored, hardening it, documenting it and shipping it as reusable ComfyUI or closed weights workflows, presets and tools other artists actually pick up
Working across both open-weight (ComfyUI, Stable Diffusion, Flux) and closed-weight (Higgsfield, Kling, Seedance) tools, benchmarking them honestly on quality, control, speed and cost
Bringing a point of view to the open vs closed debate and helping us decide where each belongs in the pipeline
Bridging the AI pipeline and our real-time work in Unreal Engine 5 where it makes sense, from engine renders as ControlNet inputs to AI output feeding back into modules
Supporting the artists who use the pipeline day to day: onboarding them, fixing what breaks and improving it based on how they really work
Holding the visual quality bar on AI output, and knowing when a frame needs a human hand before it ships
Who are you?
A technical artist with an AI pipeline background: an art eye plus pipeline thinking, not a pure engineer
You have built a generative AI production pipeline that other artists actively use in production. A one-off demo or a personal experiment doesn't count here; we want to see something a team relied on
You have trained models (LoRAs or similar) on specific characters and styles, and can talk through your dataset, captioning and training choices and what you'd do differently. You have trained skills in Claude , Chat gpt or similar that you can walk us through
Deep, hands-on ComfyUI experience: building, debugging and versioning node graphs, custom nodes, and workflows others can run without you in the room
Strong working knowledge of Stable Diffusion and Flux, ControlNets, LoRAs, IP-Adapters, upscaling and inpainting
Real production experience with closed-weight video tools such as Higgsfield, Kling and Seedance, and a clear view of where they beat open-weight tools and where they don't
A background in Unreal Engine or Unity is a strong plus.
Ideally you've made the move from real-time 3D into generative AI
Comfortable with Python scripting and a repo-based workflow of branches, pull requests and review
A real plus if you've worked on video consistency problems (character, motion, temporal coherence) rather than stills alone
What You Must Be
A builder for other people: you measure success by how many artists use what you made, not by how clever it is
Tool-agnostic and evidence-led: you'll argue for open or closed weight based on results, and change your mind when the results change
High output and high standards, able to move fast in a space that shifts every few weeks without letting quality slip
Direct and collaborative: you give and take clear feedback, and you raise problems early
Able to show a portfolio, showreel or pipeline walkthrough that proves the work is yours, including how other artists used it
The hard parts The tools you build on this quarter may be obsolete next quarter, so a lot of the job is deciding what not to adopt.
And our content is safety-critical: a frame that looks great but shows the wrong PPE or an impossible machine can't ship, so you'll be held to accuracy as well as beauty. Our process A 30-minute intro call, a technical conversation where you walk us through a pipeline you've built, a practical session on a real Pixaera brief, and a final chat with the founders. We aim to complete the whole process within 2 weeks and will always tell you when you'll next hear from us.
Benefits
How do we support and benefit our employees:
Competitive salary and stock options - everyone shares in what we're building
25 days holiday + public holidays
Fully equipped setup from day one (no waiting for laptops)
Regular company offsites and team socials
A flat, transparent culture with direct access to leadership
At Pixaera, we're committed to building a workplace where everyone feels respected, supported, and empowered to thrive. We believe that diverse perspectives fuel creativity and innovation, which is why we celebrate individuality and foster an environment of inclusion and belonging.
We're proud to be an equal opportunity employer - discrimination of any kind has no place here. No matter your background, identity, or lived experience, you're welcome at Pixaera.
Thông tin chung