Senior AI Software Engineer

About the Role

We're seeking a Senior AI Software Engineer to architect and build the core agent platform that powers enterprise autonomous workflows at AGI Dialect. You'll work on critical infrastructure that enables our agents to reason, plan, and execute complex business processes with high efficiency, reliability, and scalability.

Responsibilities

• Architect and implement production-grade agent infrastructure that powers autonomous workflows across diverse enterprise environments with high reliability and performance.
• Stay current with emerging capabilities from LLM providers and research teams, integrating new model features and providing unified interfaces for seamless interaction with different model sizes and architectures.
• Drive continuous improvement in speed and quality of the agent codebase, optimizing engineering processes to deliver exceptional product experiences consistently.
• Build robust generalized evaluation systems for agent capabilities, powering both online production metrics and offline experimentation across all product surfaces.
• Collaborate across engineering, research, product, and design to identify gaps, make strategic tradeoffs, and ship high-impact features.

Requirements

• 4+ years of industry software engineering experience.
• Demonstrated experience building agent applications utilizing tool calling, Model Context Protocol (MCP), and advanced context engineering techniques.
• Strong intuition for making architectural tradeoffs when selecting and combining different models to achieve optimal cost efficiency and system reliability.
• Self-motivated with proven ability to take ownership of platform components and drive them to completion.
• Solid understanding of distributed systems, API design, and production deployment practices.

Nice to Haves

• Experience with open-source LLM post-training, fine-tuning, or evaluation methodologies.
• Contributions to agent frameworks or AI infrastructure projects.
• Experience with prompt engineering, retrieval-augmented generation (RAG), or agentic orchestration patterns.