Machine Learning Research Engineer - Virtual Cell

SandboxAQ tests quantum-based navigation technology on Northrop Grumman UAS. Northrop Grumman's Lumberjack(R)// Image Source: SandboxAQ September 9, 2026 BY SandboxAQ SandboxAQ has completed a flight test of its AQNav magnetic navigation technology aboard Northrop Grumman's Lumberjack unmanned aircraft system, the companies said Sept. 9. SandboxAQ said the test was the first reported demonstration of a magnetic navigation, or MagNav, system on an attritable aircraft. The company also described it as the first reported pairing of magnetic navigation and visual navigation on an attritable, one-way-attack platform. The flight test involved integrating SandboxAQ's commercial, dual-use AQNav technology with Northrop Grumman's Lumberjack, a Group 3 unmanned aircraft designed for attritable operations. The companies are developing the capability for environments where GPS signals may be unavailable, disrupted or spoofed. Military forces operating unmanned aircraft increasingly face contested navigation environments in which conventional satellite-based positioning can be unreliable. AQNav uses magnetic-field measurements and physics-based models to determine an aircraft's position without depending on satellite signals or other externally transmitted navigation sources. SandboxAQ says the technology can operate passively in different environments and can complement inertial, visual and satellite-based navigation systems. The company positions AQNav as an alternative positioning, navigation and timing, or Alt-PNT, capability for aircraft operating in GPS-denied conditions. Northrop Grumman program manager Max Schuster said the integration is part of the company's effort to develop resilient navigation capabilities for autonomous and unmanned systems. The Lumberjack was designed and developed in less than 14 months from its first flight, according to Northrop Grumman. The company describes the aircraft as a modular Group 3 UAS intended for attritable missions. SandboxAQ said the flight demonstration also highlighted a new software-focused version of its AQNav platform. Unlike a system requiring dedicated navigation hardware, the software-only architecture is designed to process sensor data using existing onboard computing resources. The software applies physics-based models to sensor information in real time and generates continuous navigation data that can be incorporated into a broader positioning, navigation and timing architecture. SandboxAQ said the system supports open-architecture interfaces and is designed to operate with operationally relevant processing latency. The approach could allow manufacturers to integrate the technology into existing and future platforms without adding dedicated processing hardware. Luca Ferrara, general manager of Navigation at SandboxAQ, said the company offers two deployment options: a full-stack system integrated into a platform's architecture or a software-only implementation designed to run within existing systems. For the Lumberjack test, SandboxAQ said engineers installed the AQNav software into the aircraft's existing systems in less than an hour. The company said AQNav has been demonstrated or designed for use across environments including open water, urban areas, feature-limited terrain and GPS-denied conditions. Its passive operation is intended to make navigation less vulnerable to interference directed at satellite-based or other externally transmitted signals. The companies said the flight test demonstrates the potential for combining magnetic and visual navigation with other onboard navigation sources to improve the resilience of autonomous aircraft. SandboxAQ and Northrop Grumman are continuing work on integrating alternative navigation technologies into unmanned systems as militaries seek greater resilience in contested electromagnetic environments.

SandboxAQ open sources Switch: bring any AI agent into any team chat. PALO ALTO, Calif. - SandboxAQ today announced Switch, software for building teams of people and AI agents in the collaboration tools companies already use. Switch is open source and can be activated within minutes in any standard enterprise environment at no cost. Switch is a platform for running AI agents together in the collaboration tools where teams already work. It connects agents into shared workspaces such as Slack and Microsoft Teams, so agents and humans can collaborate in the same channel rather than handing work between separate tools. Switch is designed to be platform- and model-agnostic: teams can bring their own agents and connect them into a common workspace. The documentation guides developers through getting Switch running, connecting an agent, and starting an agent workflow in their existing collaboration environment. Developers are building increasingly capable AI agents, but putting them to work across a team still requires integrations and repeated context sharing. Switch turns channels in tools such as Slack, Microsoft Teams and Discord into shared rooms where people and agents work together with the same context, resources, and history. Through MCP, APIs and adapters, Switch is vendor agnostic, so teams can use agents and models across providers without locking their workflows, context, or collaboration environment into a single AI ecosystem. "Every team we talk to has capable agents trapped in separate tools, and a coordination tax eating the productivity those agents were supposed to deliver," said Mohammed Aboul-Magd, General Manager, SandboxAQ. "Switch pulls agents into the channels where teams already work, where knowledge compounds instead of being relearned with every task. If you can message a colleague, you can put an agent to work." Key capabilities available in Switch today include: * Open source: Quick Start and Team editions deploy in minutes, with an extensible architecture for enterprise deployment. No vendor lock-in. * Bring your own agents: Connect existing agents built with Claude Code, Google ADK, LangChain, OpenAI, and other leading frameworks * Work where teams already work: Bring agents into existing collaboration platforms including Slack, Microsoft Teams and Discord. * Keep context with the work: Rooms preserve context and history as people and agents join, leave, and hand off work. * One agent, many rooms: Use the same agent across projects and teams, with each room carrying its own context, participants, and rules. The Switch team used Switch to build Switch, with its engineering, design, and marketing teams collaborating inside rooms throughout development and launch. Switch works through rooms, or channels that form around the work itself. Engineering and security teams can pull specialized agents into a live incident with no briefing required, because the room already holds the timeline. A marketing team can staff a campaign room with research, copy, and design agents, with every asset building on the last. A legal team can add drafting and review agents that already have the deal's full context, making each round of review faster. As agents come and go, the context, history, and rules stay with the work, so the team's output compounds instead of resetting with every hand-off. Switch is part of Flint AI, SandboxAQ's portfolio of AI agent products. It extends the path from testing a single agent locally with the free Flint CLI, which developers use to scan and evaluate individual agents, to putting people and their agents to work together in shared rooms. Switch is available today on GitHub. Developers can get started with the Tutorials, explore the documentation, and learn more at FlintAI.dev. The software is released under Apache 2.0 with the Commons Clause and is free for use by anyone. Users can utilize the software within their enterprise, personally and also issue free products with the software.

SandboxAQ launches open-source Switch for human and AI agent collaboration. August 27, 2026 SandboxAQ has announced Switch, software for building teams of people and AI agents in the collaboration tools companies already use. Switch is open source and can be activated within minutes in any standard enterprise environment at no cost. Switch is a platform for running AI agents together in the collaboration tools where teams already work. It connects agents into shared workspaces such as Slack and Microsoft Teams, so agents and humans can collaborate in the same channel rather than handing work between separate tools. Switch is designed to be platform- and model-agnostic: teams can bring their own agents and connect them into a common workspace. The documentation guides developers through getting Switch running, connecting an agent, and starting an agent workflow in their existing collaboration environment. Developers are building increasingly capable AI agents, but putting them to work across a team still requires integrations and repeated context sharing. Switch turns channels in tools such as Slack, Microsoft Teams and Discord into shared rooms where people and agents work together with the same context, resources, and history. Through MCP, APIs and adapters, Switch is vendor agnostic, so teams can use agents and models across providers without locking their workflows, context, or collaboration environment into a single AI ecosystem. "Every team we talk to has capable agents trapped in separate tools, and a coordination tax eating the productivity those agents were supposed to deliver," said Mohammed Aboul-Magd, General Manager, SandboxAQ. "Switch pulls agents into the channels where teams already work, where knowledge compounds instead of being relearned with every task. If you can message a colleague, you can put an agent to work." Key capabilities available in Switch today include open-source Quick Start and Team editions that can be deployed in minutes, with an extensible architecture for enterprise deployment and no vendor lock-in. Organisations can connect existing agents built with Claude Code, Google ADK, LangChain, OpenAI and other leading frameworks, and bring these agents into collaboration platforms such as Slack, Microsoft Teams and Discord. Switch Rooms preserve context and history as people and agents join, leave and hand off work, while the same agent can be used across multiple projects and teams, with each room maintaining its own context, participants and rules. The Switch team used Switch to build Switch, with its engineering, design, and marketing teams collaborating inside rooms throughout development and launch. Switch works through rooms, or channels that form around the work itself. Engineering and security teams can pull specialized agents into a live incident with no briefing required, because the room already holds the timeline. A marketing team can staff a campaign room with research, copy, and design agents, with every asset building on the last. A legal team can add drafting and review agents that already have the deal's full context, making each round of review faster. As agents come and go, the context, history, and rules stay with the work, so the team's output compounds instead of resetting with every hand-off. Switch is part of Flint AI, SandboxAQ's portfolio of AI agent products. It extends the path from testing a single agent locally with the free Flint CLI, which developers use to scan and evaluate individual agents, to putting people and their agents to work together in shared rooms. Switch is available today on GitHub. Developers can get started with the Tutorials, explore the documentation, and learn more at FlintAI.dev. The software is released under Apache 2.0 with the Commons Clause and is free for use by anyone. Users can utilize the software within their enterprise, personally and also issue free products with the software.