Full-Time
Senior Computer Vision Engineer
Dexterity
51-200 employees
AI-powered robots autonomously load trailers
Compensation Overview
$160k - $200k/yr
San Carlos, CA, USA
In Person
 AI & Machine Learning (2) |
|---|
| , |
- BS/MS in Computer Science, Machine Learning or a related discipline, or equivalent experience
- 5 or more years of related work experience
- Experience with OpenCV and Open3D
- Prior experience building, training and deploying production ML models from scratch using PyTorch and TensorFlow
- Strong knowledge of Modern C++ and Python
- Experience using profilers and debuggers to optimize code
- Experience building and maintaining production code
- Experience with serving architectures like NVIDIA Triton, TorchServe and TensorFlow Serving, Flask and others
- Strong academic background and knowledge of ML and academic papers
- Experience with cloud infrastructure such as AWS, GCP, Azure etc. and experience using this infrastructure to create training and serving pipelines
- Ability to trace and solve problems across interconnected systems, pipelines and applications
- Experience with Git and modern CI pipelines
- Writing Computer Vision code and building Machine Learning models for computer vision
- Understanding current systems and their limitations and bringing expertise in Computer Vision and developing ML models
- Updating and scaling current ML pipelines to cover more scenarios and improve accuracy
- Leverage expertise around subfields of semantic segmentation and instance segmentation, keep up to date with latest developments and implement them using platforms like PyTorch, TensorFlow etc
- Experience with RGBD datasets and point clouds is very important to the type of problems you will encounter in this role
- Experience in ML and CV for autonomous vehicles or robotics is very useful
- Solid grounding in writing optimized code in C++ and Python
- Familiarity with cloud infrastructure to build training and serving pipelines
- End to end process of gathering, augmenting, labeling data and curating datasets, studying performance and optimizing quality, training, deploying and troubleshooting models on the field
- End to end process of gathering, augmenting, labeling data and curating datasets, studying performance and optimizing quality, training, deploying and troubleshooting models on the field
- Ability to build highly performant models and serving architectures for low latency near real-time inference
- Open source solutions understanding useful but also experience building own models and setting up training pipelines
- End to end process of gathering, augmenting, labeling data and curating datasets, studying performance and optimizing quality, training, deploying and troubleshooting models on the field
- Experience with Docker and Kubernetes
- Previous startup experience
Dexterity.ai builds AI-powered robotic systems for logistics and supply chain tasks, with a focus on trailer loading. Its robots autonomously pick parcels and place them into crossbelt or tilt-tray sorters at speeds faster than human workers. The system understands its environment in real time without needing external data feeds or warehouse software integrations, which makes deployment easier across different sites. The company sells the robotic hardware and AI software to businesses and also provides consultation services to help integrate and optimize automation. Dexterity.ai aims to improve operational efficiency and employee safety while making logistics work more attractive to workers, helping customers—from small companies to large enterprises like FedEx—achieve faster, safer, and more scalable parcel handling.
Company Size
51-200
Company Stage
Late Stage VC
Total Funding
$291.2M
Headquarters
Redwood City, California
Founded
2017
Simplify's Take
What believers are saying
- FedEx scales trailer loading across US hubs post-2026 Investor Day demo.
- Partnerships with HIWIN and Kawasaki expand hardware production and market reach.
- $50K Foresight API Challenge builds developer ecosystem for new applications.
What critics are saying
- FedEx concentration risks revenue if it shifts to Boston Dynamics in 12 months.
- Kawasaki partnership erodes IP as it builds competing software stacks in 18 months.
- OpenAI and Tesla world models surpass Foresight, collapsing moat in 24 months.
What makes Dexterity unique
- Foresight world model enables 17x speedup on NVIDIA L4 GPUs for truck loading.
- Mech dual-armed robot performs 4D box-packing in under 400ms without warehouse integration.
- Hardware-agnostic across four robot types and five hands for six logistics applications.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Flexible vacation time
Comprehensive healthcare benefits
Retirement plan
Wellbeing & fitness offerings
Insurance programs
Discounts & perks
Growth & Insights and Company News
6 month growth
↑ 0%1 year growth
↑ 0%2 year growth
↑ 0%Dexterity.ai has detailed how its production-proven world model, Foresight, is achieving significant performance improvements on NVIDIA hardware as the company scales in enterprise environments. The company showcased the technology at FedEx's 2026 Investor Day in Memphis. Foresight, built from over 100 million autonomous actions in production, gives robots real-time understanding of the physical world. Dexterity's engineering team redesigned the perception pipeline using NVIDIA L4 GPUs and TensorRT, reducing processing time from 1,508 milliseconds to 90 milliseconds per cycle — a 17-fold speedup. The system now processes 32 times more data per cycle. NVIDIA recognised the collaboration at GTC 2026. At the FedEx event, attendees watched Dexterity's dual-armed robot autonomously load a trailer. FedEx discussed plans to scale trailer loading and unloading across several US hubs.
Dexterity's Foresight world model applies physical AI to truck loading. Dexterity Inc. today announced advancements in its physical AI stack for industrial robots, anchored by Foresight, a world model and 4D box-packing agent. The company said these advances can help with some of the most physically demanding and hardest-to-staff tasks, such as truck loading. In addition, Dexterity is launching the Foresight API (Application Programming Interface) Challenge with up to $50,000 in prizes for student teams. Founded in 2017 at Stanford University's robotics lab, Dexterity said it "builds full-stack physical AI systems to address the most labor-intensive tasks in warehouse and logistics operations." The Redwood City, Calif.-based company described Foresight as "a physics-consistent world model, an agentic skill framework, and interpretable safety-first architecture." Leading enterprises around the world use Foresight for truck loading and unloading, parcel singulation, and palletizing and depalletizing, added Dexterity. Foresight models the physical world. Foresight is a transactable representation of the physical environment that enables robots to perceive, reason, and act. "Foresight represents a new class of world model, built not for observation, but for physical manipulation at the production scale," asserted Dexterity. In autonomous truck loading, Foresight powers Mech, Dexterity's dual-armed "superhumanoid" robot. The 4D box-packing agent reasons across three spatial dimensions plus time, determining where to place each package onto an evolving wall of freight. This is a combinatorial problem far more complex than the game of Go, with near-infinite input variation, up to 400 potential placements per box, and multiple walls packed simultaneously. Foresight makes each placement decision in under 400 milliseconds, explained Dexterity. The system optimizes density, stability, reachability, and dual-arm parallelism while also predicting how each placement affects the integrity of the entire truck, the company said. Architecture is interpretable, safety-first. Built on Foresight, Dexterity claimed that its agentic framework coordinates perception, decision, and motion agents that operate asynchronously to automate truck loading, package sortation, and other applications. "The architecture is interpretable and safety-first, giving operators visibility into why the system makes each decision," noted the company. Dexterity added that its physical AI stack is application-agnostic and hardware-agnostic, proven in production across six applications and a developer platform. It is running on four robot types and five hand types. To date, Foresight has been trained with experience from over 100 million autonomous actions in production. "Foresight delivers real-time, production-grade random box packing in 4D space-time, predicting how one placement dictates the integrity of the entire truck," stated Samir Menon, founder and CEO of Dexterity. "Physical AI is not just a future promise; it is a system that perceives, decides, and acts in the real world, right now." Dexterity launches Foresight API Challenge. To give the physical AI community exposure to production-grade world models, Dexterity has launched the Foresight API Challenge. Student teams build packing agents and compete on a public leaderboard for up to $50,000 in prizes. No simulator is provided; competitors must build their own understanding of the physics. Challenge details and signups are available at dexterity.ai/challenge. In addition, Dexterity said its browser-based truck-loading game lets anyone experience the problem firsthand. The company was a 2024 RBR50 Robotics Innovation Award honoree for its trailer-unloading system.
Dexterity launches Foresight world model and 4D packing agent. Physics-constrained world model delivers growth in physical ai-powered truck loading. By Robotics 24/7 Staff March 10, 2026 Foresight is Dexterity's world model, the intelligence layer that lets its robots reason about the physical world, predict what will happen next and act with confidence in environments where mistakes are expensive and safety is non-negotiable. Stay up-to-date with news and resources you need to do your job. Research industry trends, compare companies and get weekly market intelligence with Robotics 24/7. Dexterity, a provider of physical AI and robotics, said it has taken a major leap forward in its physical AI stack, anchored by Foresight, its new world model and 4D box packing agent. The company said that these advancements help solve some of the most physically demanding and hardest-to-staff tasks, such as truck loading. Alongside the announcement, Dexterity also launched the Foresight API Challenge with up to $50,000 in prizes for student teams. A world model built for the physical world. Dexterity said that ForeSight is a physics-consistent world model, generating a real-time, transactable representation of the physical environment that enables robots to perceive, reason and act. The company said that Foresight represents a new class of world model, built not for observation, but for physical manipulation at the production scale. In autonomous truck loading, Foresight powers Dexterity's Mech dual-armed "superhumanoid" robot, with a 4D box packing agent that reasons across three spatial dimensions plus time, determining where to place each package onto an evolving wall of freight. Dexterity said that this is a combinatorial problem far more complex than the game of Go, with near-infinite input variation, up to 400 potential placements per box and multiple walls packed simultaneously. Foresight makes each placement decision in under 400 milliseconds, jointly optimizing density, stability, reachability and dual-arm parallelism, while predicting how each placement affects the integrity of the entire truck. An interpretable, safety-first architecture. Built on Foresight, Dexterity's agentic framework coordinates perception, decision and motion agents that operate asynchronously to automate truck loading, package sortation and other applications. Dexterity said that the architecture is interpretable and safety-first, giving operators visibility into why the system makes each decision. "Foresight delivers real-time, production-grade random box packing in 4D space-time, predicting how one placement dictates the integrity of the entire truck," said Samir Menon, founder and CEO of Dexterity. "Physical AI is not just a future promise, it is a system that perceives, decides and acts in the real world, right now." Dexterity said that this Physical AI stack is application-agnostic and hardware-agnostic: it is proven in production across six applications and a developer platform, running on four robot types and five hand types. To date, Dexterity said that Foresight has been trained with experience from over 100 million autonomous actions in production. Foresight API Challenge. To give the Physical AI community a window into production-grade world models, Dexterity also announced the launch of the Foresight API Challenge in March. Student teams can build packing agents and compete on a public leaderboard for up to $50,000 in prizes. Dexterity said that no simulator is provided, and competitors must build their own understanding of the physics. Dexterity also launched a browser-based truck loading game that lets anyone experience the problem firsthand. Latest in perception. Latest in artificial intelligence. Latest robotics news.
Dexterity, a Physical AI robotics company, has unveiled Foresight, a state-of-the-art world model and 4D packing agent designed for autonomous truck loading. The system enables Dexterity's dual-armed robot, Mech, to make placement decisions in under 400 milliseconds whilst optimising density, stability and reachability. Foresight operates as a physics-consistent world model that allows robots to perceive, reason and act in real-time. The system handles a combinatorial problem more complex than Go, managing up to 400 potential placements per box across multiple walls simultaneously. Trained on over 100 million autonomous actions, it runs across six applications, four robot types and five hand types. Dexterity is launching the Foresight API Challenge, offering student teams up to $50,000 in prizes to develop competing packing agents using the company's API.
Also in March, Dexterity unveiled Mech and its new physical AI platform.