The Software Engineer Making Excavators Drive Themselves
There are exactly four people who decided that the right moment to automate the $13 trillion global construction industry was now, and that the right machine to start with was an excavator. Tom Eliaz is one of them.
As Co-Founder and VP of Engineering at Bedrock Robotics, Eliaz leads the team writing the software stack that turns a standard 20-to-80-ton excavator into an autonomous machine - one that digs, moves, and grades a construction site without a human in the cab. Not a prototype. Not a demo. A deployed system that, in late 2025, moved 65,000 cubic yards of dirt across a 130-acre manufacturing site in the American Southwest while the rest of the industry watched.
Investors noticed. In February 2026, Bedrock Robotics closed a $270 million Series B led by CapitalG and the Valor Atreides AI Fund, with NVIDIA's venture arm, MIT's investment office, and Tishman Speyer among the backers. The company reached a $1.75 billion valuation. Total funding: over $350 million. Not bad for a startup that went public less than eight months earlier.
Eliaz's path to autonomous construction equipment is not a straight line - it's a series of bets on machines that connect to the internet in unusual ways. First it was cloud infrastructure for toy robots. Then enterprise data pipelines for a company that Salesforce eventually bought for $3.2 billion. Then leading 100+ engineers building customer engagement software at scale. Then, in 2024, a sharp left turn into autonomous heavy machinery.
Robotics Funding
Valuation (2026)
at Bedrock
Moved Autonomously
From Penn Engineering to the Cloud to the Construction Site
Eliaz graduated from the University of Pennsylvania with a BSE in Computer Science and Engineering in 2002. He spent his early career building systems that didn't get much press but mattered: a healthcare technology company focused on patients with chronic conditions, then an early engineering role at Anki - a robotics startup in San Francisco making consumer toys with actual onboard AI.
At Anki, Eliaz wasn't working on the robots themselves. He was building the cloud. Starting in 2012, he stood up infrastructure from scratch capable of handling millions of connected devices - the backend nervous system that let Cozmo and Vector communicate, update, and learn. It's the kind of problem that sounds mundane until you try to do it at scale, with real-time responsiveness, for millions of households. He stayed seven years.
When Anki shut down in 2019, Eliaz joined Segment, a customer data infrastructure company that had already established itself as essential plumbing for digital businesses. He ran engineering for the NYC office and led a range of infrastructure and product areas. When Twilio acquired Segment for $3.2 billion, Eliaz stayed on, eventually becoming VP of Engineering for Twilio Engage - a role that put him in charge of over 100 engineers building AI-driven customer engagement products at one of the largest communications platforms in the world.
"I can't contain in a post the impact the last five years have had on me... I'm excited to be diving into my next adventure with some old and new friends."- Tom Eliaz, LinkedIn post announcing his departure from Twilio, 2024
In 2024, Eliaz left Twilio. His LinkedIn farewell post deliberately said nothing about what came next - just a hint about "old and new friends." Months later, Bedrock Robotics would emerge from stealth with $80 million and a clear mission: deploy autonomous systems on construction sites using the same class of technology that made Waymo cars possible.
The Bedrock Bet: Retrofitting, Not Replacing
The Bedrock Operator is not a new machine. It's a retrofit system - LiDAR sensors, RTK-GPS, high-definition cameras, and an industrial computer - that bolts onto any standard excavator in the 20-to-80-ton range and gives it centimeter-level situational awareness. The software stack processes the sensor fusion data in real time, translates a contractor's grading plan uploaded to Bedrock's Fleet Portal, and drives the machine through the task without a human at the controls.
The business logic is clean: contractors already own excavators. Nobody wants to buy a new $500,000 machine. Bedrock's hybrid model combines a hardware installation fee with a monthly software subscription. The excavator keeps working 24/7 - pausing only for refueling and maintenance. For large infrastructure projects with impossible labor timelines, that's not a feature. It's the whole point.
Eliaz brings an unusual credential to this problem. Unlike the Waymo veterans who form the core of Bedrock's co-founding team - CEO Boris Sofman, CTO Kevin Peterson, and VP Engineering Ajay Gummalla all worked on autonomous driving - Eliaz's background is in building the invisible systems that keep connected machines running at scale. He's the engineer who already did this for millions of toy robots. He knows what breaks when you go from lab to production.
Bedrock Operator - Autonomous Excavation System
A retrofit kit that turns any standard 20-80 ton excavator into an autonomous machine. Contractors upload grading plans to the Fleet Portal. The machine handles the rest.
- LiDAR Sensors
- RTK-GPS Navigation
- High-Definition Cameras
- Sensor Fusion AI
- 24/7 Operation
- Fleet Portal Software
- Real-Time Data Processing
- Human Supervisor Override
From Stealth to $1.75B in Under a Year
Seed + Series A announced July 2025 as combined $80M raise. Series B announced February 2026.
Twenty-Two Years of Building Systems That Run at Scale
Software engineering careers are often measured in languages learned and frameworks shipped. Eliaz's career is more usefully measured in categories of connected machine: first medical devices and patient data, then consumer robots, then enterprise data pipelines, then customer engagement AI, and now construction equipment.
Each move brought a bigger fleet and harder constraints. Anki's devices were consumer-grade and needed to work in millions of living rooms. Segment's pipelines processed hundreds of billions of events per day. Twilio Engage served some of the world's largest enterprise customers with real-time AI. Bedrock's machines operate in construction sites - unstructured, dangerous, and as far from a controlled environment as it gets.
The thread running through all of it is infrastructure thinking: build systems that stay up, handle edge cases, and scale without heroics. That's what Eliaz brings to a co-founding team stacked with autonomous vehicle engineers. He's not the robotics person. He's the person who makes the robotics work in production.
Bedrock Robotics Investor Roster
What Comes After the Excavator
Bedrock's roadmap makes clear that the excavator is the entry point, not the destination. The company is targeting fully operator-less deployments in 2026 - the year this article publishes - and plans to expand to dozers, loaders, and articulated trucks. The vision is a fully connected, orchestrated fleet: multiple machine types, multiple tasks, running continuously on a single large-scale construction site, coordinated through software.
That's a very large distributed systems problem with physical consequences. It's also, in a compressed form, the problem Eliaz spent two decades learning to solve.
Construction has a labor shortage that no amount of recruitment will fix fast enough. Demand for housing, data centers, energy infrastructure, and factories is outpacing the available workforce of skilled equipment operators by a margin that's getting wider every year. The industry needs a different answer. Bedrock Robotics is betting it's autonomous machines, running around the clock, managed through software rather than crew schedules.
Tom Eliaz is writing that software.