AI Engineer

Mission: Design, develop and optimize the artificial intelligence infrastructure powering Julius 3.3 and future i.Wonder technologies.

  • Founding-Team Interest
  • Remote (United States)
  • 0.60% proposed equity
  • No visa sponsorship

Founding-Team Interest — Equity-Based Opportunity. i.Wonder Laboratories is currently identifying prospective founding-team members. Cash salaries have not yet been established. These listings are expressions of interest in future positions, not offers to begin unpaid employment. Any work arrangement will require a legally compliant agreement. Future team members are expected to make a serious time commitment once employment arrangements are finalized.

Apply online All roles

The role

Design, develop and optimize the artificial intelligence infrastructure powering Julius 3.3 and future i.Wonder technologies. Build advanced retrieval systems, contextual intelligence, memory infrastructure, model evaluation frameworks and inference optimization systems that improve the accuracy, reliability and capabilities of our Compound AI Agents.

Agentic AI Engineer and AI Engineer

Agentic AI Engineer builds the mechanisms that let Julius plan, coordinate agents, execute tools and complete missions. AI Engineer builds the underlying intelligence, retrieval, memory, model and evaluation systems that make Julius more capable.

See also: Agentic AI Engineer.

Key responsibilities

  1. Advanced RAG systems: Develop codebase-aware retrieval using vector embeddings, AST-based semantic code chunking, hybrid search, reranking and knowledge graph retrieval.
  2. Memory & Project Brain: Build intelligent retrieval across semantic, episodic and procedural memory. Integrate Project Brain so Julius can access project architecture, requirements, research, previous decisions and mission history.
  3. Model selection & inference: Optimize model routing, inference performance, latency, context usage, provider costs and model reliability.
  4. Evaluations & benchmarking: Develop AI evaluation datasets and systems measuring retrieval quality, coding accuracy, reasoning, hallucinations, groundedness and model performance. Maintain separation from independent acceptance reviews.
  5. Context engineering: Improve prompt construction, information prioritization, context compression and knowledge delivery to Julius's different AI modules.
  6. Model adaptation: Research fine-tuning, distillation and specialized models when justified by evaluation results and available training data.
  7. Multimodal intelligence: Help Julius interpret visual environments, screenshots, voice, audio and structured world-state information as these capabilities mature.

Immediate engineering priorities

  • Advanced codebase-aware RAG
  • Project Brain retrieval and memory
  • AI evaluation infrastructure
  • Context and inference optimization

About i.Wonder

i.Wonder Laboratories, Inc. builds multiplayer open-world AI creation platforms, and researches, develops and deploys Compound AI Systems and Compound AI Agents. Our thesis is that AI creation should not end when a chat window closes: people and Compound AI Agents should be able to build, collaborate, compete, create software and maintain what they make, together, over time, inside a persistent world.

Wonderland is our flagship platform: a persistent multiplayer open world for AI creation, designed so that AI-powered creation and real software development happen in the same place as collaboration, competition and play. Wonderland is in active development.

Julius is our flagship Compound AI Agent and the intelligence inside Wonderland. It governs three permanent tiers — a Prompt Architect (a large language model), a Coding Agent (an agentic coding system) and an Independent Reviewer (an independent agent harness) — and every tier reads from and writes to Project Brain, Julius’s durable memory. Work is finished when independent review and verification say so, not when an agent says it is.

Read the research papers · Watch the Wonderland simulation

What to know about this stage

i.Wonder is an early-stage company. Wonderland and Julius are in active development, and the descriptions of them here are their intended design, not a report of finished features. The research papers set out the architecture and the evidence still to be gathered.

How to apply

Apply online with the application form. You’ll upload your résumé (PDF or DOCX), add your LinkedIn profile and links to relevant work — for example shipped projects, repositories, a portfolio or writing — and answer two short questions about this role.