Job description
What you’ll build:
● An agentic orchestration layer for memory, context-switching, and real-time intent.
● Retrieval-augmented generation (RAG) pipelines for grounded, high-fidelity responses.
● Prompting systems optimized for fast, emotionally intelligent conversations.
● Tools for prompt versioning, observability, and testing.
● Custom fine-tuning for LLMs and eval loops that prioritize character fidelity.
● Tight collaboration with video/audio ML, infra, and design teams to bring it all together.
● Have shipped LLM-powered, multi-agent, or RAG-based systems in production.
● Are comfortable with LangGraph, CrewAI, or building orchestration layers from scratch.
● Understand prompt engineering deeply — and know its limits.
● Have experience evaluating and fine-tuning models for responsiveness and style.
● Care about identity, memory, and coherence in human-AI interaction.
● Want to help define what it feels like to speak to an AI.
● Salary $120,000 (USD) per year plus equity options available on top.
● Platinum health insurance — we cover the good stuff.
● 4% 401(k) match — long-term thinking encouraged.
● Flexible hybrid schedule in SF Bay Area, or remote for the right person.
● We’re pre-seed, so base comp is lean — but the founding equity package and impact are real.
● Early ownership in a team that values velocity and thoughtfulness.
● AI Confidant included — weird, personal, powerful.
