Build the stack behind AI presence.
Wonders.AI is a small team in Taipei building the Darlin LLM, the Kotodama Model and the Darlin tech stack. Six core-team roles are open, each working directly with the founder.
We want people who are passionate about waifus, anime and interactive characters, and who live it every day.
Open roles
Core team · six positionsLead research and training of our proprietary models: the Darlin LLM and Kotodama. Hands-on research leadership working directly with the founder on architecture, training and evaluation.
- Develop, train, debug and optimise our LLM, TTS, ASR and speech models
- Pretraining, continued pretraining, SFT, LoRA/PEFT, reinforcement learning and distributed GPU training
- Improve personality consistency, emotional response, multilingual dialogue and character behaviour
- Advance Kotodama: emotional TTS, ASR, voice cloning and low-latency speech
- Build multilingual datasets, training pipelines, benchmarks and automated evaluation
- Improve RAG, long-term memory, quantisation, KV caching and on-device inference
- Proven experience training and improving real model weights, not only API integration
- Strong PyTorch, Transformers, Python and GPU training expertise
- Deep LLM or speech AI specialisation with cross-domain range
- Independent research, debugging and production deployment experience
- A genuine love of anime, waifus and expressive AI characters
Own the Unity architecture of our character platform across Windows, macOS, iOS, Android and XR, and make characters feel alive in real time.
- Own Unity architecture, features, debugging and optimisation across PC and mobile
- Animation controllers, IK, facial expressions, gestures and real-time behaviour
- Integrate LLM, TTS, ASR and streaming responses into seamless character interaction
- Build MR/AR experiences and cross-device interaction for emerging XR platforms
- Optimise CPU/GPU, memory, rendering and responsiveness across hardware
- Strong Unity/C# expertise with shipped commercial products
- Real-time 3D, animation systems, profiling and optimisation
- Strong architecture, debugging and independent problem-solving
- AI integration, mobile, networking or XR experience valued
- A genuine passion for anime, waifus and interactive characters
Own the entire product experience across PC, mobile and web: audit, redesign and see improvements through to implementation.
- Lead, audit and redesign the product experience on every platform
- Design voice, conversation, character, memory and real-time AI interactions
- Onboarding, navigation, settings, payments, notifications and edge cases
- Figma prototypes within a week, production-ready designs within a month
- Design systems and consistent components across platforms
- Work directly with engineers and review the real implementation through release
- Proven experience designing and shipping AI-powered products
- Exceptional UX, interaction design and product judgment
- Advanced Figma, prototyping and design-system expertise
- Portfolio of real products with measurable UX improvements
- Five or more years; real enthusiasm for anime, games and Japanese entertainment
Own the visual direction and technical art quality of our characters: anime-style rendering, shaders, animation and performance, directing and implementing yourself.
- Set the visual identity and quality standards; guide artists
- Refine faces, eyes, hair, skin, clothing, expressions and lighting
- Unity URP, NiloToon, PBR, Shader Graph, HLSL, materials and post-processing
- Rigging, blendshapes, IK, facial expression and natural movement
- Hands-on modelling, textures, VFX, lighting and scene composition
- Optimise shaders, pipelines and assets for PC and mobile; build Unity/C# art tools
- Portfolio with exceptional anime aesthetics and technical execution
- Professional Unity, shader, 3D production and rendering experience
- Can set art direction and still work hands-on
- Ownership, independent problem-solving and obsessive attention to detail
Own and scale operations: launches, community and customer experience, campaigns, and the internal systems that make a small team execute. Works directly with the founder.
- Launch planning, operational readiness, incident escalation and post-launch reviews
- Japan, Taiwan and global community operations; support workflows, FAQs and feedback loops
- Release coordination: testing, user-facing flow checks and fix coordination
- Campaigns, creator and IP collaborations, events and user acquisition; crowdfunding and pre-orders when needed
- SOPs, accountability, task tracking, metrics and efficient workflows
- Coordinate product, engineering, design, marketing and support to remove blockers
- Proven experience leading launches, operations or growth campaigns
- Hands-on community, customer support and operational experience
- SOPs, KPIs and cross-functional delivery
- Chinese and English required; Japanese strongly preferred
- Interest in anime, gaming, AI characters and Japanese entertainment
Turn the product vision into launch-ready experiences: requirements, quality, release management and execution across engineering, design, research and operations.
- Translate vision into requirements, priorities, user flows and acceptance criteria
- Personally evaluate usability, AI interactions, onboarding, payments, character behaviour, voice and memory
- Own testing, bug tracking, fix verification and release readiness
- Coordinate engineers, designers, researchers and operations; resolve blockers
- Analyse community feedback, support issues and product data
- Improve activation, onboarding, engagement, retention and conversion
- Proven experience shipping consumer software or AI products
- Exceptional product taste, UX judgment and attention to detail
- Hands-on QA, debugging and release discipline
- Chinese and English required; Japanese strongly preferred
- Interest in anime, AI characters and Japanese entertainment
Apply
Tell us which role and send a link to your work. Suitable applicants hear back within a working day. Applicants take a short company assessment.
Or email contact@wonders.ai directly.