Description
Summary We are developing a suite of AI agents for 3D modeling, interior design, architecture, and procurement. These agents power customer-facing experiences while also interacting with our internal systems and automating the workflows behind that experience. The work is not a generic chatbot project. We need agents that can reason over product data, call tools, coordinate across services, retain shared memory, and produce reliable outputs that can be used in production workflows. This is an existing product and codebase. You will work with our team to design, implement, test, and improve agents that are useful, observable, and safe to deploy. Scope of Work You will help build and improve agents that interact with our product, design, and project systems. The initial areas include: - SKU ingestion agent: gathers and validates data for new products in our system, enriches SKU attributes, identifies missing information, and prepares product records for downstream use. - Design agent: selects products based on a user's input, room context, design preferences, constraints, and available catalog data. - Project agent: keeps users informed via text or voice, tracks scope and budget, coordinates project status, and supports procurement workflows. The work will include: - Designing workflows, tools, prompts, state, and evals. - Implementing APIs, databases, and customer-facing surfaces. - Building reliable tool-calling, retries, and handoffs. - Improving product-data quality, normalization, and retrieval. - Shipping production code, tests, logging, and docs. - Measuring behavior with evals, traces, and regressions. Technical Expectations We expect this work to involve using TypeScript, Python, and Mastra to create backend services, customer-facing applications, and data systems. You should be comfortable working across application code, agent orchestration, APIs, and databases. Qualifications - Strong full-stack engineering background with production systems experience. - Experience building AI agents, tool-using LLMs, or workflow automation systems. - Strong TypeScript experience, ideally including agent frameworks such as Mastra. - Python experience for data processing, service integration, or model-adjacent workflows. - Experience integrating LLMs with real databases, APIs, queues, customer-facing workflows, or internal tools. - Comfort with structured product data, catalogs, SKUs, variants, and data quality. - Ability to design evals, trace behavior, debug failures, and improve reliability over time. - Good judgment around human-in-the-loop workflows, guardrails, and permissions. - Clear written communication and ability to explain tradeoffs and decisions. - AWS, PostgreSQL, and voice/SMS integration experience are nice to have.