SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Kinter builds AI agents for mid-market finance teams, automating repetitive, judgment-heavy work like prepaid accounting, accruals, payroll, and reconciliation. The company is a lean, product-centric team of ~30 people that launched its core product a few months ago.
In this role, you will evolve the AI harness powering Kinter's agents, focusing on orchestration, tool-calling, context and memory management, and the scaffolding between models and production accounting workflows. You'll design and ship new agentic capabilities, from novel workflows to structured interfaces that enable agents to act reliably within financial systems of record.
Key responsibilities include:
- Strengthening human-in-the-loop patterns, confidence thresholds, and fallback behaviors to ensure agents prioritize correctness over fluency (e.g., escalating when uncertain rather than confidently guessing).
- Expanding evaluation and observability tooling: curated datasets, evals, automated regression tests, tracing and logging to understand agent behavior at scale.
- Working directly with customers and their finance teams to determine where agents can act autonomously versus where human oversight is required, translating that judgment into the harness.
- Helping define Kinter's agentic architecture as it expands from expense management to a broader set of agents.
- Collaborating closely with product, engineering, leadership, and customers; decisions move fast and your work is highly visible.
The role emphasizes operating in a sensitive domain where correctness is paramount. You'll work across the full stack, from AI orchestration to web services, in a startup environment where the roadmap shifts with customer needs.
REQUIREMENTS:
- 2+ years of hands-on experience building and shipping autonomous agentic systems in production (not prototypes or demos).
- 8+ years of experience building and shipping SaaS products.
- Experience operating agents in sensitive domains (finance, healthcare, legal, or similar) with strong instinct for why correctness and human-in-the-loop patterns matter.
- Direct, hands-on experience with agent frameworks (LangGraph, AgentCore, AutoGen, CrewAI, LlamaIndex) tied to a shipped product.
- Solid experience building web applications and services with Node.js and TypeScript; comfort working across the stack.
- Excellent problem-solving skills: diagnosing issues with incomplete information, breaking down ambiguous problems, building resilient systems around probabilistic components.
- Generalist and self-starter mindset; energized by learning new domains (accounting) and owning features end-to-end.
- Strong collaboration skills across cross-functional teams; knowledge-sharing, pairing, removing bottlenecks.
- Experience running production systems (deploying, testing, monitoring, debugging) with reliability and observability built in from the start.
- Comfort with ambiguity and startup pace.
