Full-Stack Machine Learning Engineer

You'll work at the heart of Sonzai Engine — the mind layer that gives AI agents persistent memory, evolving personality traits, and per-interaction learning. This is a full-stack ML role: you'll design neural architectures, build training pipelines, and ship production APIs that developers integrate directly.

The engine processes behavioral signals in real time — extracting personality dimensions (OCEAN model), building hierarchical memory structures, and updating agent state with every conversation turn. Your work will directly shape how millions of AI agents understand and relate to the humans they interact with.

• Behavioral Profiling Pipeline — Real-time extraction and update of OCEAN personality dimensions, communication style markers, and preference signals from conversation data.
• Hierarchical Memory System — Multi-tier memory architecture (episodic, semantic, procedural) with efficient retrieval, consolidation, and forgetting mechanisms.
• Per-Interaction Learning — Online learning systems that update agent behavior after every message without catastrophic forgetting. Mood detection, habit formation, relationship depth scoring.
• Embedding & Retrieval Infrastructure — Vector stores, similarity search, and context assembly for agent memory recall at inference time.
• Production APIs — TypeScript/Python services powering the SDK, MCP integration, and REST API that developers use to create and interact with compounding agents.

• Strong ML fundamentals — transformers, embeddings, online learning, representation learning.
• Production ML experience — model serving, feature pipelines, monitoring, and iterating on models in production.
• Full-stack capability — comfortable building APIs (Node/Python), working with databases (Postgres, vector stores), and deploying infrastructure.
• NLP / conversational AI experience — understanding of dialogue systems, sentiment analysis, personality modeling, or memory-augmented architectures.
• Systems thinking — you can reason about how behavioral signals flow from raw conversation data through extraction, storage, retrieval, and back into agent responses.

• Experience with personality computing, affective computing, or computational social science.
• Familiarity with the OCEAN/Big Five personality model or similar psychometric frameworks.
• Contributions to open-source ML tooling or published research in relevant areas.
• Experience building developer-facing APIs or SDKs.