Timur Rollins
I lead creative production, and I build the AI systems that run it.
Seven years of professional delivery, four of them producing high-end motion and 3D for global brands. Alongside it I design and build AI systems and put them to work in real production, from the first brief to final delivery.
Barcelona
Production
Seven years delivering motion, 3D and branded content — four of them running the work end to end for global brands. Client footage stays with the clients; the record is below.
Creative Project Manager
ContentLabs / TransPerfect · Apr 2024 – Jul 2026
Owned scope, schedules, stakeholder review and final delivery across multiple parallel productions. Managed multidisciplinary teams of motion designers, 3D artists, sound specialists and external vendors, and folded AI-assisted steps into the pipeline without loosening the quality gates.
Clients Epic Games · Amazon · Microsoft · Pfizer · AbbVie · LVMH · Hermès · LinkedIn
Motion Designer & Producer
We Are Love Int. · Oct 2023 – Apr 2024
Produced and delivered branded content end to end for marketing campaigns: scripts, storyboards and visual direction, plus live-stream production scheduling, talent logistics and technical handoffs against broadcast timelines.
3D Designer & Animator
Freelance · Aug 2022 – Sep 2023
Client projects end to end — scoping, timelines, revisions and delivery — producing 3D assets and animation for cinematic and advertising use, including Unreal Engine 5 work with MetaHuman and motion capture.
Selected work
Four projects, all mine. An AI creative engine, two sci-fi worlds I write and draw, and a brand I co-founded. Some of it is code, some of it I draw by hand.
AI creative engine
Emily
Designed and built solo, end to end · 16 episodes published
A pixel-cat character with her own voice, look and running series. It began as one pipeline I wrote: a data-driven animation engine, a text-to-speech voice that runs on my own machine, chiptune sound, and a one-command render. I hand it a topic, it hands me back a finished, captioned video ready to post.
It is now a crew. Nine specialist agents each own one stage of the process, hand their work to the next as a file, and can send it back if it isn't good enough. Emily runs the floor. I only step in at two decisions: approve the story, approve the final cut.
▸ How it's built
One character, one JSON, one render
▸ Play with the cast · interactive
Every character, face, hairstyle and move is drawn in code — no image files. Click around, toggle animations, swap a backdrop and press play.
Original sci-fi IP · creator
Life Was Purrrfect
Story, character design, concept art and panels
A dark science-fiction story of mine: a cat rebuilt as a machine, alone on a derelict ship. I designed the world and the characters and drew it, running an AI-assisted pipeline that takes a rough sketch through concept, lighting and finish to a final panel. It stays my hand and my eye, just faster.
▸ How it's built
Storyboard
Concept sketches
Moodboard & development
Character & vehicle design
Process: the AI-assisted pipeline
Brand · creative direction · co-founder
Brand identity, art direction and launch campaign
A sculptural-décor brand I co-founded and art-direct, built around one signature object: a hand-finished alabaster egg in numbered editions. I shaped the whole visual world: the identity, the product concept, the launch film, and the store itself. The campaign follows one egg through warm, lived-in rooms, then glowing as a light after dark.
▸ How it's built · the film, campaign & pipeline
The launch film
The launch campaign
Process: how the campaign was made
The campaign imagery is generated rather than photographed. I direct the pipeline, then grade and cut the results.
Animated series · degree thesis · creator
Concept, claymation, 3D scanning, environments and animation
A post-apocalyptic Barcelona reclaimed by mutated nature, told as a dark, tender animated series. A lone wanderer and his cat move through a city where Gaudí's architecture is overtaken by stone-statue birds, mechanical rats and bioluminescent plants that heal and spread. Every character began as real clay on my desk: I sculpted them, 3D-scanned them, and brought the scans into the engine to light and animate. Made as my degree thesis and shown at IED Barcelona.
▸ How it's built
Posters
Storyboard
Process
Concept
Frames
Systems I run
Five things I designed, built and run myself. One is a product you can open and use. The rest are groups of small programs that read live data, act on it, keep an eye on each other, and restart themselves when something breaks.
Live product · you can try it
AgentUp
live on the web · 27 tests, green CI · built solo in two days
A training app for contact-centre agents. You practise against an AI customer who does not make it easy, by chat or by voice, and then you get scored. The design decision I care about is that there are two separate grades. One judges how good the call was, out of a hundred. The other is a straight pass or fail on the rules, because a warm, friendly call that breaks a regulation still has to fail. Those are two different questions and most tools blur them into one number.
Open the live app
Automation at scale
Market automation stack
15 processes · ran on a regulated exchange · months of unattended uptime
A group of programs that read live market data, act on rules, and watch each other. One process spawns and streams the rest, a risk engine holds hard limits, and a separate watchdog sits on top as a dead-man's switch that can stop everything at once. It checked its own state against the exchange every cycle, restarted anything that crashed within seconds, and ran for months without me touching it. The trading logic was never the hard part. The hard part was building something I could trust to run unattended, with hard limits it could not talk itself out of.
AI job search
The apply engine
1,000+ roles sourced & scored · a tailored CV and cover letter for every one · 185 applications submitted
An agent that runs my whole job hunt. It pulls live roles from 40+ company boards, scores each one against my real profile with a language model, writes a CV and cover letter for that exact role, fills in the application, answers the screening questions, and stops for me to check before anything is sent. It runs on a schedule, and it will not claim experience I don't have. The CV on this page came out of it.
Private AI assistant
Odysseus / Emily
100% local · reads my other systems through a read-only mirror · cites its sources
A research assistant that runs entirely on my own machine. Local language models, a vector memory, nothing leaves the building. Odysseus is an open-source local-AI setup PewDiePie put out. I took it, upgraded it, and pointed it at my own systems. Emily is the face I gave her, the same character from the content engine above. She reads everything I run through a read-only mirror and gives me a short, sourced brief every morning. She can only read it, never change it, which was the whole point.
Deployed inside a studio
NuniBot
shipped into a live review pipeline · comments-only, never touches the team's data
A production assistant I built for a creative studio's review-and-approval pipeline. I built her to take the chasing off our art director. She catches media links that get lost in review comments, pings the right producer, and keeps each project's job number and status up to date. One rule she never breaks: she can comment freely, but she can only write two named fields, and every other write throws. That constraint is what made her safe to run against live production data. I gave her a face and a voice so people would actually want to use her. Say hi, she'll remind you to do your Notion.
▸ NuniBot, in detail · case study
The problem, the safety model, and what she is allowed to touch
The problem
A studio runs its reviews in Frame.io and its projects in Notion, and neither knows the other exists. Client comments land on a review link and never reach the project page. Review URLs go missing inside other fields. Projects reach approval with the reporting fields still empty. Producers chase all of it by hand, and the studio's reporting ends up full of holes.
What she does
A client comments on a Frame.io review, and NuniBot posts that comment on the matching Notion page and mentions the producer, routing it to the right sub-item or parent project. If a project has an empty Frame link field while the URL is sitting in another field, the page body or a comment, she fills it in. When a client project reaches Job Approved with reporting fields missing, she nudges the producer instead of guessing. Once a day she posts a health brief: deadlines near or overdue, projects with no producer or no job number.
The safety model, which is the actual work
She runs against live client projects, so the interesting engineering is not what she can do. It is what she cannot. Every write in the codebase goes through one file. Comments are allowed. Property writes are allowed only for names on an explicit whitelist, currently Frame link and Job Number, and any other property write throws. No function anywhere can change a status, a value, a client or a date, edit page content, touch the database schema, or delete anything. A dry-run mode logs every intended write and performs none, so a change can be watched before it is trusted.
That constraint is what made her safe to point at a working studio's real data on day one.
Where the taste comes from.
Four years producing motion, 3D and branded content for gaming, cinematic and advertising clients, owning delivery from brief to final asset, on top of seven years of professional delivery. Real-time work in Unreal Engine 5, MetaHuman and motion capture. That craft is why the systems I build ship work that actually looks right, and why I care about the last ten percent.
About
AI is the biggest lever regular people have ever been handed. I'm here to make sure you use it before it gets used on you.
I'm a producer who builds the automation. I design and build AI and automation systems and run them in real production, and I've led delivery on twelve to fifteen productions at once for global brands. So I build the technology and I manage the work around it. That combination is my edge.
After four years producing motion, 3D and branded content, I bring AI into real workflows to make them faster and more reliable, and I turn manual, repetitive work into systems that run on their own. I would rather ship a rough thing that works than plan a perfect thing that doesn't. Everything on this page got built and shipped.
- Based in
- Barcelona, open to relocation
- Work authorization
- US citizen + Swiss passport. US, EU/EEA & Switzerland
- Languages
- English C2 · French C2 · Spanish C1
- Works with
- Node, Playwright, local LLMs, GPT, Unreal Engine, Adobe CC
- Currently
- Open to roles in AI-assisted production, applied AI, implementation and delivery, and creative operations
