Founding AI/ML Engineer
@ Guthrie AI
Construction is one of the world's largest industries — and one of the last to see meaningful software innovation. Subcontractors have been running the same manual estimating workflows for 15+ years. Guthrie AI is changing that.
Guthrie builds AI-powered Virtual Bid Assistants that help subcontractors organize RFPs, manage RFIs, and generate proposals — so estimators can spend less time on paperwork and more time winning profitable work. The platform is in production, trusted by leading subcontractors, and showing real results.
Just closed a seed round (term sheets signed). ~$2M ARR on the core product with a plan to 10x in 12 months. A team of 9 engineers — 2 AI/ML, 5 full-stack, and a Chief AI Officer — operating the platform alongside real customers every day.
Construction diagrams and floor plans are not in the training data of any off-the-shelf foundation model. OpenAI, Anthropic, Google — none of them work well out of the box on this material. That's not a bug, it's a moat.
Guthrie's bet is that the team that can train and iterate on their own computer vision models — segmentation, object detection, classifiers for floor plans and technical drawings — will own the AI layer for this industry. Off-the-shelf models have hit their ceiling here. This role exists to push past it.
Founding AI/ML Engineer, reporting to the Chief AI Officer. The 2–3 existing ML engineers will look to you for technical direction. Principal-level IC / technical lead — not a people manager, but leadership matters.
Computer Vision — the irreplaceable part
Segmentation models (SAM-tier, 2D depth), object detection, and classifiers on floor plans and construction drawings. Architectural model improvements — not just fine-tuning.
Agentic Orchestration & Retrieval
LLM fine-tuning, RAG pipelines, local vs. foundation model trade-offs, cost optimization. Important, but the team can survive with "good enough" if the CV depth is exceptional.
- Train and improve in-house CV models for construction diagrams (segmentation, object detection, classifiers)
- Lead architectural and model-improvement work where off-the-shelf models hit their ceiling
- Own the model development roadmap and set technical direction for the AI/ML team
- Build agentic orchestration and retrieval flows (RAG, fine-tuning, local vs. foundation model decisions)
- Provide technical leadership and mentorship to 2–3 existing AI/ML engineers
Strong Signal
- Deep 2D segmentation depth — SAM, SAM2, SAM3 or comparable
- Python + PyTorch + CUDA fluency with production systems shipped or strong CV publication record
- Has trained models from scratch when off-the-shelf hit their ceiling
- Published work in raster-to-sequence, floor plan localization, or structural priors
- Exposure to construction diagrams, floor plans, or architectural drawings (extremely rare — gold)
- Prior technical lead or mentorship of ML engineers
Not a Fit
- TensorFlow-only without PyTorch transition
- Pure backend ML without CV depth
- Off-the-shelf orchestrator who can't train custom models
- Fine-tuning APIs only — no architectural model work
East Coast hybrid — 3 days/week at the Philly hub (Amtrak-friendly from NYC and DC). Flex for exceptional candidates.
Reply to the note that brought you here, or grab 15 minutes on my calendar — happy to tell you more about the team and the technical challenges before you decide anything.