Applied AI Engineer

Paradigm is a software company transforming the way that the residential construction & building product industries operate across the globe. We are looking for an Applied AI Engineer to be part of revolutionizing these industries. The Applied AI Engineer will design and deploy AI systems that power next-generation construction workflows. In this role, you’ll build the bridge between machine learning models and real-world applications—using LLMs, computer vision, retrieval systems, and agentic orchestration to automate key steps in the homebuilding process such as plan interpretation, takeoffs, estimating, and specification matching. This role blends hands-on engineering, AI system design, and integration with enterprise platforms—turning AI capabilities into production-ready features that builders, designers, and estimators can rely on daily. What You Will Do: Build and deploy agentic AI systems that combine large language models, computer vision, and rules-based logic to automate construction workflows. Implement retrieval-augmented generation (RAG) pipelines to connect LLMs with real-time project data, specifications, and plan sets. Integrate AI models into production services and APIs for use in design, estimating, and procurement applications. Fine-tune and adapt pre-trained models for domain-specific tasks such as plan understanding, entity extraction, and bill-of-material generation. Measure and improve model accuracy, reasoning quality, and cost efficiency through Evals, zero/few-shot tests, Chain-of-Thought reasoning analysis, and LLM-as-a-judge evaluation techniques. Optimize inference performance and monitor system reliability in live environments. Work closely with data engineers to structure, index, and prepare multimodal datasets (text, CAD/BIM files, images) for AI consumption. Implement robust data pipelines for training, validation, and feedback loops to continuously improve performance. Ensure reproducibility, observability, and version control across model artifacts and pipelines. Collaborate with product and software engineering teams to embed AI outputs into digital platforms, ERP systems, and supplier integrations. Build APIs and SDKs that make AI features accessible to design and estimating applications. Ensure security, compliance, and data integrity across integrations. Partner with Applied ML Engineers and Software Engineers to move from prototype to production in fast, iterative cycles. Participate in internal experiments and contribute to evaluations and model improvement efforts. Stay current with emerging AI frameworks, agentic orchestration techniques, and developer tools. What You Need to Succeed: Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Machine Learning, or a related field is preferred. 5+ years of experience designing, building, and deploying AI or ML-powered systems in production, with recent hands-on work applying LLMs or agentic architectures. Hands-on experience with LLMs, retrieval-augmented generation (RAG) pipelines, and multimodal models from providers such as OpenAI, Anthropic, or Hugging Face. Proficiency in Python and at least one major ML framework (PyTorch, TensorFlow). Strong familiarity with agentic orchestration frameworks (e.g., LangGraph, Temporal, n8n, or similar) to design multi-step, tool-using AI systems. Experience deploying and integrating AI models using enterprise AI platforms such as Azure AI Foundry, AWS Bedrock, or Google Vertex AI. Working knowledge of retrieval systems, vector databases, and semantic search for contextual grounding. Understanding of prompt engineering, model evaluation, and cost/performance optimization for large-scale inference. Excellent problem-solving skills, curiosity, and a collaborative mindset. Exposure to construction or design automation workflows is preferred. Experience working with plan files, CAD/BIM data, or digital twin systems is preferred. Background in evaluation-driven AI development, including zero/few-shot testing, Chain-of-Thought reasoning, and LLM-as-a-judge validation is preferred. Familiarity with data labeling, feedback loops, and fine-tuning to improve domain-specific accuracy is preferred. Familiarity with agent performance metrics and lifecycle management across production environments is preferred.