Staff Engineer, Applied ML

Circuit•Austin, TX

About The Position

As a Staff Engineer, Applied ML at Circuit, you will join Product Engineering to build machine learning capabilities for the product. You will work closely with Forward Deployed Engineers (FDEs), Product Engineers, and design customers to understand real-world challenges and turn them into robust, scalable solutions. You will identify when a customer need points to a broader product capability and build it so it can serve multiple customers and use cases. This is a hands-on engineering role focused on current product and customer problems. You will use machine learning advances where useful, exercise judgment about what is ready to ship, and own solutions end to end, with attention to accuracy, reliability, latency, and cost. The work draws primarily on LLMs, alongside traditional neural networks and statistical models, and includes understanding customer use cases and product interactions, modeling data and processes, building on existing systems, evaluation, production implementation, and model training or fine-tuning where beneficial. You will work end to end with design customers, build shared capabilities such as Circuit’s evaluation framework, and improve how product quality is measured and regressions are detected. Product Engineering builds the core platform and works directly with customers and design partners. FDE teams own solutions architecture and technical demonstrations; you will collaborate with them on what to build and partner with Core ML engineers when problems call for deeper work on document models or the agent loop. Circuit’s production product includes a long-running agent loop written in Go, driven by frontier model APIs and supported by specialized tools for retrieval, extraction, and agent execution. The company uses open-source and internally fine-tuned models, particularly vision-language models, hosted on KServe. Customer data includes technical documents, high-resolution blueprints, raster images, vector geometry, CAD, and structured data.

Requirements

Track record of owning and shipping production systems end to end, with strong software engineering fundamentals.
Deep experience with data structures and data modeling across relational and graph databases, including well-reasoned opinions on schema design, normalization, and indexing strategy.
Hands-on experience building with LLMs, RAG systems, and agent harnesses, including a solid grounding in their operation, failure modes, and evaluation.
Experience working directly with customers or field teams and translating what is learned into product development.
Track record of generalizing a solution built for one customer so others can deploy it.
Experience building and maintaining features across a multi-service architecture.
Strong proficiency in languages such as Python and Go.
Fluent daily use of AI coding tools, including writing effective prompts, iterating on incorrect output, and distinguishing working code from code that only appears correct.
Ownership mindset focused on delivering working systems in production.
Strong product judgment and understanding of how system behavior affects user experience and trust.
Ability to rigorously review agent instructions, skills, and documentation for clarity, lack of ambiguity, and reliability.
Curiosity about how systems behave in practice and how customers use them.
Bias toward shipping and ability to drive ambiguous problems to working solutions without a fully defined path.
Judgment about what is ready to ship and willingness to push back to keep work grounded in customer outcomes.
Clear communication, including explaining machine learning tradeoffs to FDEs and customers in plain terms and sharing progress early.
Ability to critically evaluate AI output and detect hallucinated APIs, incorrect logic, and subtle bugs.
Systems thinking, including attention to correctness and failure modes.
Low ego and focus on team outcomes.

Nice To Haves

Expertise in retrieval systems, including embeddings and ranking.
Expertise in production model training and fine-tuning across LLMs, deep learning, or traditional machine learning.
Expertise in evaluation design and dataset curation from messy, real-world customer data.
Expertise in debugging production machine learning systems with AI observability tools.
Experience with vision and multimodal models, especially for technical documents such as PDFs and CAD containing text, vector geometry, and raster content.
Experience in construction, civil engineering, or other industrial domains.
Experience with open-source models such as Qwen or DeepSeek and serving them with vLLM, KServe, or similar.
Hands-on PyTorch or similar experience for model training and fine-tuning.

Responsibilities

Work end to end with FDEs and design customers to build Circuit solutions that address customer challenges.
Generalize and package solutions and methods so FDEs can efficiently deploy them across customers.
Improve underlying models, harnesses, evaluation methodologies, and related capabilities.
Debug production systems, including edge cases and failure modes in machine learning pipelines and retrieval systems.
Build and maintain evaluation systems, including datasets, scoring approaches, and repeatable regression testing.
Model customer data and processes so agents can reason over materials such as spreadsheets, PDFs, and product rules.
Improve retrieval quality, relevance, or latency for customer-specific use cases.
Trace production issues to their root causes and fix them in the product rather than relying on one-off patches.