Software Engineer (AI Agents)

Akur8•New York, NY

23h•$170,000 - $190,000•Hybrid

About The Position

Akur8 is hiring a Software Engineer for its AI team, building an end-to-end platform that transforms large volumes of insurance regulatory and market documents into searchable, trustworthy intelligence embedded in pricing workflows. The role involves applied AI systems and production engineering, with work spanning agent systems, evaluation and data quality, and reliable engineering at scale. Engineers own the product feedback loop: identifying shortcomings, investigating root causes, developing evaluations, and shipping improvements. The role also includes direct customer interaction when useful for understanding the problems being solved. Projects may include designing tool-calling and agent-orchestration patterns for multi-step research across related documents; building ingestion and normalization systems for millions of complex, versioned filings; developing reinforcement fine-tuning harnesses and graders for research tasks with non-verifiable rewards; measuring and optimizing agent pipeline latency, cost, and quality; and running experiments through to product recommendations. The position is hybrid in New York City and requires three days per week in the office.

Requirements

2–5 years of professional software development experience and strong programming fundamentals.
Proficiency with LLM APIs, agent frameworks, and modern retrieval patterns, and an interest in keeping current with the field.
A strong track record of shipping high-impact work in ambiguous situations, including defining success, delivering an initial version, and iterating.
Experience shipping user-facing products, in backend or full-stack roles.
Interest in building rigorous evaluation infrastructure to determine whether changes improve results.
Strong written and spoken English.
Strong interpersonal skills and ability to work as a team player.

Responsibilities

Design and refine tool-calling patterns and agent orchestration for multi-step research tasks across large sets of related documents.
Build ingestion and normalization systems for complex structured documents at scale.
Build and iterate on reinforcement fine-tuning harnesses, including training graders for research tasks with non-verifiable rewards.
Develop tools to measure, profile, and optimize agent pipeline performance across latency, cost, and quality.
Run experiments end to end, analyze results, and translate findings into product recommendations.
Diagnose product shortcomings, analyze root causes, build evaluations, and ship improvements.
Work directly with customers when it helps develop a deeper understanding of their problems.