Technical Architect

Role overview Lead the technical architecture for a production platform that helps construction teams work with drawings, specifications, and bid documents. The role spans a product application, machine learning and computer vision services, AI agents, and data infrastructure; it combines long-term system design with hands-on engineering in a fast-shipping environment.

Responsibilities - Own architecture across application, ML/CV services, agent runtime, and data platform, balancing customer needs with delivery speed. - Design for scale, reliability, and cost as document and drawing processing grows substantially. - Set direction for model infrastructure, orchestration, evaluation, registries, and compute budgets. - Lead design reviews, code reviews, and technical documentation; mentor senior engineers on system-wide decisions. - Partner with product and executive leadership on a 12–18 month roadmap, including build, buy, and deprecation choices. - Continue to write and review production code, contribute to incident response, and tackle difficult technical problems.

Requirements - 10+ years building production software, including 3+ years in a senior staff, principal, architect, or CTO role at a product company. - Deep experience in at least two areas: large Rails or Python applications, production computer vision, large-scale LLM or agent systems, or data-intensive B2B SaaS. - Hands-on cloud experience with AWS and/or GCP, PostgreSQL at scale, background job systems such as RabbitMQ or Sidekiq, and modern deployment practices. - Evidence of architectural decisions that supported 10× or greater growth. - Recent production coding experience and intent to remain hands-on. - Familiarity with reliability and DevOps practices, including Kubernetes, Helm, and CI/CD, and ability to work with SRE colleagues.

Nice to have - Experience in construction, architecture, engineering, or other vertical SaaS, or helping a company scale technically from Series A to C. - Experience shipping systems that handle unstructured real-world inputs such as PDFs, images, and OCR, and operating production CV or agent systems with evaluation practices.