Kantiv (formerly Joist AI) - Senior AI Engineer - Chat & Agent Systems

• Approximately 4–5 years of professional software engineering experience. • Strong proficiency in Python, including writing idiomatic, typed, testable, and maintainable production code. • Strong programming fundamentals and consistently sound engineering judgment. • Experience designing, delivering, and operating production systems. • An ability to create useful abstractions while keeping systems simple and understandable. • Strong knowledge of API design, data modeling, concurrency, error handling, and observability. • Experience writing effective automated tests with tools such as pytest. • The ability to review code beyond surface-level concerns and explain the reasoning behind suggested changes. • Experience mentoring junior developers and improving the quality of their work. • Comfort taking ownership, identifying opportunities, and driving improvements across a team. • Clear written and verbal communication. • Hands-on experience building LLM or agentic applications is required. Relevant experience may include: • Building tool-calling or multi-step workflows using LangGraph or a comparable agent framework. • Designing conversation state, context management, and execution flows. • Working with model APIs, structured outputs, retrieval, and grounding. • Supporting streaming responses and asynchronous execution. • Evaluating prompts, models, and end-to-end agent behavior. • Instrumenting and debugging LLM applications with Langfuse, LangSmith, or similar observability platforms. • Managing the reliability, latency, and cost of production LLM systems. • We do not expect expertise in every named tool. We care about the underlying engineering judgment and the ability to learn or replace frameworks as the ecosystem changes. • The candidate should be a power user of agentic coding workflows and understand how to use coding agents as engineering tools rather than simple code generators. • They should be able to: • Use coding agents effectively across exploration, planning, implementation, testing, debugging, and review. • Provide agents with the right context, constraints, and verification steps. • Critically evaluate generated code for correctness, security, maintainability, and unnecessary complexity. • Design development workflows that keep engineers accountable for the resulting code. • Identify repeatable team workflows that can be improved through agents and automation. • Teach junior developers how to use coding agents effectively without weakening their engineering fundamentals. • Lead practical initiatives that make the team faster and more AI-native. • WHAT SUCCESS LOOKS LIKE • The team makes clearer and more consistent technical decisions. • Projects have simple designs, well-defined boundaries, and useful tests. • Code reviews catch meaningful design and correctness issues early. • Junior developers receive actionable guidance and grow more independent. • The team develops effective and responsible agentic coding practices. • Production issues become easier to diagnose and resolve. • The codebase becomes easier to understand, change, and operate over time.