Now Reading
Tech Behind Construction Analytics Company Doxel

Tech Behind Construction Analytics Company Doxel

  • Doxel has raised millions of dollars in venture capital from investors such as Andreessen Horowitz and has customers like Kaiser Permanente and Royal Dutch Shell.

Founded in 2016 by Saurabh Ladha (CEO) & Robin Singh (CTO), Doxel is an artificial intelligence company building an AI-Powered Project Controls solution for the construction industry. 

An encounter with a customer asking, “telling me what is happening today is good, but come back when you can tell me what is going to happen next,” left an everlasting imprint in the minds of the founders. “This was a game-changer for us and started the next three years of product development, customer feedback, iteration, and ultimately large scale customer acquisition – the sole driver of which is our AI-Powered Project Controls platform which can tell owners and general contractors where they are today, objectively, and exactly where they will be in the future in terms of cost and schedule on their project,” said Saurabh Ladha, CEO, Doxel.

Register for our Workshop on How To Start Your Career In Data Science?

Plans for India

The company has raised $56.5 million to date in funding. With the funds, the company is hiring across all functions, including doubling the team in India. Saurabh said, “Our India head is one of our engineering leaders and foundational to our technological success. The team in India brings deep technological expertise and construction expertise, especially in Building Information Modeling (BIM). Doxel will continue to invest in this team which is pivotal to Doxel’s success. Additionally, our fully integrated India/US workforce has a blended culture which ultimately underpins our success.”

Their USP

Doxel is the only solution in the market that can take video footage of a job site, 3D designs, project budget, and project schedule, and tell clients exactly how much progress has occurred on their job site today and how much will occur in a week, a month, or a year,” said Saurabh. 

This available data can apprise clients about any further delay in the project, along with a quick estimate of how much money this delay will cost. This predictive analysis ability is unique to Doxel in construction and enables clients to mitigate the challenges proactively. As per Saurabh, Doxel has some specific capabilities which we don’t see in the market today:

  • Track all trades and systems
  • Track any project type
  • Track 100% of the site
  • Can turn around critical insights every 72 hours
  • Predictions are >96% accurate after the second scan

Tech Behind

The company employs a wide range of ML models of different complexity, ranging from simple heuristics to Graph Neural Networks (GNN), specially architected to compare mesh and point cloud data. 

“GNNs are very well suited for handling the sparse and irregular nature of point cloud data. The GNN we employ learns local features that relate points to their neighbours in the scanned point cloud and global features describing the broader shape of the cloud. Thereafter, these features are represented by layers that act on the graph of nearby points in an analogous manner to convolutional operators acting on pixels in images. Other models project the cloud and mesh into 2D along various axes, allowing us to leverage the many methods of deep learning on images to better model our 3D data,” said Robin Singh, CTO & Co-founder, Doxel.

“We work at the intersection of Building Information Models, ML-based Computer Vision, big data pipelines, and business analytics; we use a large spectrum of tools to build the end-to-end stack,” he added.

See Also

Tech Stack

  • For BIM and CAD, we have built custom plugins on top of standard construction software (like Autodesk NavisWorks). 
  • Use managed services from standard cloud providers to power backend flow (k8s, database, data pipelines, security, etc.). 
  • The company’s persistent data is stored at secured cloud storage across multiple services. 
  • Spark and Airflow are used for batch processing data pipelines. 
  • For ML, Tensorflow and PyTorch frameworks for writing deep learning models.  
  • Use of advanced VSLAM techniques to localise panoramas to enable accurate object detection in 2D. 
  • The frontend is powered by React and uses REST APIs to communicate with backend infrastructure.

Hiring at Doxel

On the hiring process, Saurabh said, “We explore multiple avenues to source candidates and leverage various support groups for diverse hiring. We are also exploring internship programs where we can work with early-stage talent and build a pipeline of future team members. In hiring, we look for performance-driven individuals who embody values of ownership, decisiveness, humbleness, empathy, and intellectual honesty.”

Road Ahead

The pandemic was a catalyst for the business, as owners and general contractors look for solutions that can improve project monitoring capabilities and support remote work. As companies looked to limit job site visits and staff on projects due to social distancing, Doxel was able to step in and provide the needed capabilities to do this type of work. 

With good funding rounds in the recent past, the company plans to continue developing its platform, hire talented staff, and better serve customers. “New customers continue to reach out to us with interest, and existing customers’ average spends continually increase as we’re brought onto more of their projects,” concluded Saurabh.

Subscribe to our Newsletter

Get the latest updates and relevant offers by sharing your email.
Join our Telegram Group. Be part of an engaging community

Copyright Analytics India Magazine Pvt Ltd

Scroll To Top