Senior Computer Vision / Applied AI Engineer - Construction Plans

KONSTRUCTIQ, INC.
Los Altos, CA, United States
1 day ago
Apply on www.disabledperson.com
Prepare application

Role details

Contract type
Permanent contract
Employment type
Full-time (> 32 hours)
Experience level
Expert
Experience required
3 years minimum
Working hours
Regular working hours

Tech stack

Computer-Aided Design Artificial Intelligence Computer Vision Building Information Modeling Software Debugging Python (Programming Language) Machine Learning Language Modeling Object Detection OpenCV Parsing Tensorflow
+4 more
Computational Geometry Pytorch Large Language Models Machine Learning Operations

Job description

We’re looking for a Senior Computer Vision / Applied AI Engineer to improve how KonstructIQ reads construction plans and generates takeoffs - detecting rooms, walls, doors, windows, fixtures, symbols, and dimensions, and turning them into reliable quantities and measurements.

You’ll be part of a small team that owns the full pipeline: document ingestion, image and vector processing, scale and geometry, detection and segmentation, multimodal reasoning, measurement, validation, and evaluation.

What You’ll Be Doing

  • Build and improve computer vision and multimodal AI systems that detect, classify, segment, and measure plan elements - walls, rooms, doors, windows, fixtures, finishes, and symbols - extracting counts, dimensions, and areas
  • Improve scale detection and geometric reasoning across drawings of varying scale, orientation, resolution, and format, combining vector PDF data, raster imagery, OCR, and multimodal foundation models
  • Decide when to use traditional CV, specialized models, multimodal LLMs, deterministic algorithms, or hybrid approaches, and build structured, production-ready outputs from foundation models
  • Build deterministic post-processing and validation systems - including confidence scoring - that turn probabilistic model outputs into stable, explainable takeoffs and flag ambiguous results instead of silently guessing
  • Build evaluation frameworks and datasets (labeling strategies, estimator feedback loops) that measure detection accuracy, measurement variance, and end-to-end takeoff quality, and turn production failures into new test cases and model improvements
  • Build visual overlays and debugging tools that make it easy to see what the system detected, measured, and missed
  • Partner with construction estimators, product, and engineering to translate real-world estimating workflows into technical solutions, and evaluate new CV/multimodal models where they materially improve accuracy or simplify the system

Requirements

  • Bachelor’s degree with 3+ years of experience in computer vision, machine learning, applied AI, or related engineering roles; MS/PhD with a focus in computer vision preferred
  • Strong experience building production computer vision systems, not just research prototypes
  • Deep knowledge of object detection, segmentation, image processing, and geometric reasoning (coordinate systems, transformations, polygons, spatial relationships)
  • Strong Python experience with CV/ML frameworks such as PyTorch, OpenCV, YOLO, Detectron2, or similar
  • Experience with multimodal vision-language models, and judgment for where foundation models outperform - or underperform - traditional CV approaches
  • Experience processing complex documents, diagrams, engineering drawings, PDFs, or other spatial/visual documents, ideally combining OCR with visual information
  • Experience designing datasets, labeling strategies, training pipelines, and evaluation frameworks
  • Strong debugging intuition - able to distinguish model, data, geometry, and pipeline problems, and optimize for real-world accuracy and consistency, not just benchmarks
  • Comfortable with ambiguous problems with no off-the-shelf solution; high ownership and high agency from experimentation through production

Bonus Points

  • Experience with architectural drawings, blueprints, CAD, BIM, construction documents, or takeoff software
  • Experience with vector PDF parsing, CAD formats, or computational geometry libraries
  • Experience with floor-plan understanding, document AI, geospatial imagery, or other precise spatial-reasoning applications
  • Experience building human-in-the-loop ML systems where user corrections improve future model performance
  • Background in construction tech or prior startup experience

Benefits & conditions

  • Act like an owner
  • Strive for excellence
  • Communicate openly and honestly
  • Lead with data
  • Succeed in work and life

Why You’ll Love Working Here

  • Work on a technically difficult computer vision problem with immediate real-world impact
  • Own a core part of KonstructIQ’s AI platform and product differentiation
  • Work with a growing dataset of real construction plans, takeoffs, estimates, and estimator feedback
  • See your models used directly by contractors to bid and execute real construction projects
  • Small team, high ownership, and the ability to shape the technical architecture from the ground up
  • Competitive salary and meaningful equity

About the company

Hey, we’re KonstructIQ

KonstructIQ is a fast-growing startup on a mission to modernize residential construction. We’re building the system of record for how contractors run their business - spanning estimating, projects, finances, and money movement.

AI is at the center of our product. Contractors can upload construction plans, photos, or describe a project and use KonstructIQ to generate scopes of work, takeoffs, and detailed estimates in minutes.

Apply for this position

This job is hosted externally. Click below to view the full posting and apply.

Apply on www.disabledperson.com
Prepare application

Good distractions

Talks and stories from around this role — technically off-topic, practically not.

3:14 min

Structuring career paths and localized data architectures

Ulrich Wurstbauer +1 · LIVE

2:35 min

Preventing remote code execution in PyTorch models

Balázs Kiss · World Congress 2023

2:56 min

Open-sourcing a complex parsing library for game data

Johan Hutting Johan Hutting · World Congress 2024

4:53 min

Achieving real-time tracking performance with OpenCV and segmentation

Thomas Endres Thomas Endres +2 · World Congress 2021

2:37 min

Optimizing technical profiles for AI sourcing and recruitment

Mina Golesorkhi Mina Golesorkhi · World Congress 2026 Europe

1:06 min

Compiling PyTorch environments for advanced time forecasting

Christoph Lohrmann Christoph Lohrmann +1 · World Congress 2026 Europe

Videos

See all

Related articles

See all