Software Development Engineer

We are looking for a Software Development Engineer to build out the core Dextr platform. This is a hybrid craft: you will design and ship traditional software services, APIs, integrations, and data pipelines, and you will engineer LLM-powered systems where prompts, evals, and model behavior are first-class parts of the stack, not an afterthought. You will own features end to end in a small, fast-moving team where your work goes in front of real hotel operators and their guests.

Responsibilities:

• Build the platform. Design, develop, and ship backend services, APIs, and integrations that power Dextr OS and our agent suite across voice, chat, and staff-facing surfaces.
• Engineer with LLMs. Build agentic workflows using LLMs as core components for prompt design, tool use, context management, retrieval, and structured outputs, and tune them against real production traffic.
• Make AI reliable. Develop evals, guardrails, and monitoring so agent behavior is measurable and accuracy improves from release to release, in line with our SmartEvals approach.
• Integrate with the hospitality stack. Connect Dextr to property systems (PMS, telephony, reservations, and staff tools) with production-grade error handling and observability.
• Own it end to end. Take features from ambiguous requirements to production independently, scoping, building, testing, deploying, and iterating based on customer feedback.
• Think in systems. Understand how your components fit into the broader platform and product roadmap, and make architectural choices that hold up as we scale.

Requirements:

• 5-8 years of professional software engineering experience building and operating production systems.
• Strong fundamentals: data structures, system design, API design, databases, and distributed systems.
• Proficiency in at least one modern backend language (e. g., Python, TypeScript/Node.js, or Go) and comfort working across the stack when needed.
• Hands-on experience building with LLMs: prompt engineering, function/tool calling, RAG, and evaluating model output quality.
• A builder's bias: you work independently, unblock yourself, and ship without waiting for detailed specs
• Product sense: you understand why you are building something and how it serves the customer, not just how.
• Clear written and verbal communication in a fast-moving, collaborative environment.