United States onlyThe Project: We are developing a highly complex, stateful, multi-agent simulation engine driven by generative AI. This is a greenfield project exploring new forms of non-linear, narrative-based user interaction at scale. Unlike typical stateless web apps, this system requires persistent "memory" for thousands of unique, evolving user sessions. We are building an architecture where multiple AI agents orchestrate complex logic, maintain long-term context, and react to user inputs in real-time.
The Role: You will serve as the Sr. AI Engineer, focusing on the core intelligence and agentic logic that drives the application. You will be responsible for designing the multi-agent architecture that manages our core interactive loops, dynamic scenario generation, and global system state aggregation. You will be the team’s primary authority on how we talk to large language models, ensuring that our agents are fast, reliable, and strictly scoped to their specific domains to minimize latency and hallucination.
This is a part-time contract role (20 hrs/week) from 4/13–5/29, with a strong likelihood of extension through October at full-time hours.
What You'll Do
Agent Design & Orchestration: Build and manage the logic for complex multi-agent workflows. You will design the systems that handle user onboarding (profile generation), dynamic scenario creation, and real-time interactive simulation loops.
Context Engineering: Architect state management for the LLMs to prevent "context rot" and hallucination. You will strictly govern what each agent knows, structuring context dynamically to maximize token caching and minimize latency.
Advanced Prompting & Evals Infrastructure: Write, test, and version-control robust system instructions for standalone LLMs and multi-agent workflows. You will design, implement, and own a rigorous evaluations (evals) framework to programmatically score both individual prompt performance and end-to-end agent lifecycles. You will establish the CI/CD-style testing loops required to iterate on model behavior predictably and safely at scale.
Moderation and Security Risk Mitigation: Design and implement pipelines in collaboration with our backend team that moderate harmful or offensive user inputs while also mitigating prompt injection attacks and undesired LLM outputs.
Full-Stack Integration: Work closely with backend and frontend teams to seamlessly integrate AI outputs into the user interface, ensuring smooth data flow from the models down to the client.
What You'll Bring
Core Engineering Foundation: Strong traditional programming background. You must understand software architecture and be capable of writing production-grade code. You cannot rely solely on AI coding assistants or vibe coding.
Generative AI Experience: 1–3 years of deep, hands-on experience building and deploying LLM-backed applications in production.
Language Proficiency: Strong proficiency in Python. Strong proficiency in TypeScript and familiarity with modern reactive frontend frameworks (preferably Angular v21).
Agent Frameworks: Hands-on experience with modern agent harnesses (e.g., LangGraph, CrewAI, OpenAI Agents SDK, Claude Agent SDK, Google ADK). Strong preference for candidates with experience using ADK.
Context & Latency Optimization: Deep understanding of how LLMs process information. You must have proven experience optimizing token usage, leveraging caching, prompt and context engineering, and designing systems that fetch only the exact context an agent needs at any given moment.
Risk Mitigation: Hands-on experience with designing and employing guardrails for agents’ actions and outputs while also mitigating prompt injection attacks.
Ideally You Are
A Pioneer: You thrive in an emerging tech landscape where best practices are still being written, and you are excited to help define them.
A Precision Communicator: You understand that a single ambiguous word in a system prompt can derail an entire multi-agent workflow.
Latency-Obsessed: You don't just care that the model gets the right answer; you care about how many milliseconds it took to generate it, and you actively design to reduce that overhead, especially when combined with content moderation and prompt injection mitigation strategies.
Core Tech Skills
Android
Augmented Reality / Virtual Reality (AR/VR)
AWS
Back-end
Creative Technologist
Database
Dev-Ops
Django
E-Commerce
Front-end
Full-stack *
GCP *
iOS
Java
Javascript *
Leader
Marketing
Media
Mobile
Node
Objective-C
Product Development
Product Manager
Prototyping *
Python *
QA
React
Swift
Systems Architecture *
Tech Producer
Typescript *
Unity
UX *
Wagtail
Prompt Engineering
Context Engineering
LLM APIs
LLM Agents
Agent Orchestration Harness (e.g. OpenAI Agents SDK / Claude Agent SDK / Google ADK / LangGraph / CrewAI)
AI Evals
AI security, guardrails, and risk management
Pay Range
- The expected pay range for this role is $61 -$78 per hour based on the US 3 pay range for a W-2 temporary engagement
- Our company has three regional pay bands that it adheres to depending on your location, we reference them as US 1, US 2, and US 3
- US 3 is our base pay. Examples of cities in US 3 are Portland, Houston and Miami.
- US 2 pay is 7.5% higher than US 3 to meet the market rates. Examples of cities in US 2 are Los Angeles, Chicago and Seattle
- US 1 pay is 15% higher than US 3 to meet market rates. Examples of cities in US 1 are Brooklyn and San Francisco
- If you are curious which region you are in, please apply and get connected with our recruiting team!
AU, CA + 5 more