Senior Software Engineer - Chat & Agent Systems

About the role

We are seeking a Senior Software Engineer to lead the design and delivery of our chat and agentic systems. The role combines hands‑on Python development with system‑level thinking to build reliable, observable, and cost‑effective AI‑driven applications.

Key responsibilities

• Own technical design and delivery of chat‑team projects.
• Shape engineering priorities, standards, and day‑to‑day technical decisions.
• Translate product requirements into simple, maintainable system designs and production Python code.
• Design clear abstractions that reduce complexity without over‑engineering.
• Review pull requests for correctness, maintainability, test coverage, and design quality.
• Mentor junior developers and improve their programming and system‑design judgment.
• Diagnose production issues across code, agent workflows, prompts, models, data, and infrastructure.
• Improve reliability, observability, latency, and cost of chat systems.
• Advance AI‑native practices by integrating coding agents throughout the development lifecycle.

Required profile

• Approximately 4–5 years of professional software engineering experience.
• Strong proficiency in Python with idiomatic, typed, testable, and maintainable production code.
• Solid programming fundamentals and sound engineering judgment.
• Experience designing, delivering, and operating production systems.
• Ability to create useful abstractions while keeping systems simple.
• Experience mentoring junior developers.
• Clear written and verbal communication skills.
• Hands‑on experience building LLM or agentic applications.

Required skills

• Python
• pytest
• LangGraph (or comparable agent framework)
• Langfuse / LangSmith (observability platforms)
• API design
• Data modeling
• Concurrency
• Error handling
• Observability
• Large Language Models (LLM)
• Agentic systems
• Tool‑calling and multi‑step workflows
• Model APIs, structured outputs, retrieval, grounding
• Streaming responses and asynchronous execution
• Prompt and model evaluation