Computer Vision Engineer

CivilGrid is the trusted civil design platform powering affordable infrastructure renewal and growth. We give utilities, civil engineers, and municipalities a single, collaborative source of truth for what's underground so teams can plan, design, and build faster, smarter, and safer.

Our platform consolidates utility records, geotechnical and environmental data, paving moratoriums, and jurisdictional requirements directly from 3,000+ utilities, agencies, and municipalities with full chain-of-custody documentation. Teams at PG&E, Mark Thomas, and dozens of leading engineering firms use CivilGrid to cut research time by up to 88%, reduce project costs, prevent dig-ins, and accelerate delivery on the projects that keep communities running.

We're a venture-backed Series A startup tackling a $4.7T US infrastructure market that has been underserved by software for decades. Founded by industry insiders, each with 10+ years in utilities and civil engineering, we move fast, ship work that matters, and stay close to the customers and field teams we build for.

As a small team, we all wear many hats and enjoy tackling new challenges. When it comes to tech, we're seeking experienced individuals who can efficiently identify problems and devise solutions. As a Computer Vision Engineer, you will get the chance to work on challenging problems that directly contribute to the scalability of CivilGrid's products, and be one of the first engineers shaping CivilGrid's computer vision capabilities.

This role reports to the Head of Product & Engineering. It is a hybrid role with expectations that you are working from our Boston, MA office at least 2 days per week if hired.

• →

  Explore our data and workflows to identify opportunities where CV/ML-based solutions can have the most impact

• →

  Apply classical computer vision techniques — feature matching, geometric transformations, image registration — to build internal and customer-facing features

• →

  Prepare training data, and train, fine-tune, and validate vision-based models

• →

  Develop cross-technology solutions that combine VLMs, vision-based models, and classical CV to solve problems no single approach can

• →

  Ship CV pipelines that can be continuously monitored and validated, with evaluation harnesses and quality metrics built in from day one

• 3 years of experience building computer vision systems in Python, with tools such as OpenCV, NumPy, and PyTorch (or equivalents)

• Solid grounding in classical computer vision: feature extraction and matching, geometric transformations, image registration, and robust estimation (e.g., RANSAC)

• Solid grounding in ML and ML-based computer vision: CNNs, transformers, embedding spaces

• Hands-on experience preparing datasets and training, fine-tuning, and evaluating vision-based models

• Start Up Ready: autonomous, moves fast, takes initiative without being asked

• Strong verbal and written communication, able to articulate ideas clearly and collaborate effectively

• A degree in Computer Science, Electrical Engineering, or a related field (B.S. or M.S.)

• Experience working with messy real-world imagery — scanned documents, OCR outputs, low-quality rasters — is a huge plus

• Familiarity with geospatial tooling (GDAL/rasterio, coordinate reference systems)

• Experience integrating VLM/LLM APIs into production workflows

• Experience with MLOps

We're a small, but rapidly growing team that ships. Expect direct feedback, real ownership from day one, and proximity to customers — huge utilities, public agencies, and engineering firms whose work depends on what we build. We default to in-person collaboration where it matters and remote-friendly flexibility otherwise. We hire people who care deeply about their craft and the communities our customers serve.