AI Engineer - Agent

AI Engineer – Agent Systems

We (https://www.kaon.io/) build production-grade LLM-based Agents for personalized content generation and long-term interactive experiences for millions of users every day. This role owns the full Agent development lifecycle — from system design and prototyping to large-scale production deployment and continuous optimization.

Responsibilities

• Design and implement production-ready Agent runtime frameworks, including orchestration engines, tool-use pipelines, memory systems, and context management modules.

• Ensure system stability, scalability, and low-latency performance under high-concurrency workloads.

• Implement asynchronous background workflows for memory consolidation and resource scheduling.

• Design secure tool-use frameworks for interacting with external APIs and services, including sandboxing, permission management, tracing, and error handling.

• Build end-to-end observability systems (logging, tracing, monitoring) and evaluation pipelines covering personalization quality, memory accuracy, and multi-turn consistency.

• Establish release workflows including canary deployment, A/B testing, and production feedback loops to drive continuous improvement.

• Optimize inference and serving performance, including caching, batching, streaming, efficient context window usage, and task scheduling.

• Build data pipelines for collecting, cleaning, and managing Agent training data, supporting SFT, RL, and reward model iteration.

• Collaborate closely with the algorithms team to translate research advances into scalable production systems.

Requirements

• Strong software engineering skills with proficiency in Python; experience building large-scale distributed systems. Familiarity with Go, Rust, or C++ preferred.

• Hands-on experience developing and deploying LLM-based Agent systems in production environments.

• Strong system design ability with sound judgment across reliability, latency, and cost trade-offs.

• Experience with asynchronous systems, task queues, and high-concurrency services.

• Deep interest in Agent memory systems, personalization, and continual learning, with the ability to bridge research and engineering.

• High ownership mindset with the ability to drive projects end-to-end from design to continuous iteration.

Nice to Have

• Experience with vector databases, retrieval systems, and RAG architecture design.

• Experience optimizing LLM inference and serving (e.g., vLLM, TGI, TensorRT-LLM) in high-concurrency environments.

• Experience building Agent evaluation frameworks, simulated interaction environments, or automated benchmarking systems.

Compensation: $140,000 – $300,000 total compensation (base + equity), depending on experience and impact.