AI systems, robotics and industrial software.
Big Bear Engineering builds and integrates software for systems that have to work outside the demo: industrial automation, multi-agent infrastructure, on-device AI, robotics and connected products.
Our work combines long experience in embedded and industrial engineering with current AI tooling. We are interested in useful systems, clear interfaces, traceable operation and engineering that can be maintained after launch.
Current systems
Robot Ross
A cyber-physical AI system combining a Huenit robot arm, model reasoning, locally hosted models and a G-code control path. The same system is used to test logging, documentation and human oversight in a real hardware loop.
Sovereign Mind
Private document question-answering for field and industrial teams. Retrieval and inference run locally, with cited answers and support for offline operation.
Canis
A forward-only experiment for reading model disposition from smartphone-class open models. Runs locally on iPhone and explores confidence, uncertainty, refusal and evasiveness before generation.
Industrial engineering
Komax — digital services
Program management for cloud services, predictive maintenance and digital products around industrial wire-processing machinery.
OPC UA Companion Specification 40570
Led the international working group defining the OPC UA Companion Specification for wire-harness manufacturing.
Audi / Komax — production integration
Data communication and integration work for wire-processing equipment on the automotive production floor.
Klepsydra Technologies
Program management for ESA certification work around an embedded AI inference engine for high-performance imaging in space.
digitalSTROM · BridgeCo · ESEC · ETH Zürich
R&D leadership in home automation and embedded networking; semiconductor-equipment software; and robotics research at ETH Zürich. Teams of up to 30 engineers and R&D programs up to CHF 5M.
What we work on
Industrial software
Machine integration, MES, cloud-connected equipment, production data and software architecture for long-lived systems.
AI systems engineering
Agent workflows, model integration, retrieval, local inference, evaluation and operational controls.
Robotics & cyber-physical systems
Model-to-machine interfaces, control paths, logging, operator interfaces and hardware/software integration.
On-device & local AI
Open-weight models on Apple Silicon and mobile devices where privacy, latency or data sovereignty matter.
Technical governance
Traceability, operational logging, documentation, human oversight and evidence generation for AI systems.
Engineering leadership
Architecture reviews, technical program leadership, team structure, integration planning and delivery recovery.
Multi-agent infrastructure, treated as an engineering system.
Flotilla is the open-source coordination layer developed at Big Bear Engineering. It exists because ad-hoc collections of AI chats do not behave like reliable engineering teams.
The framework provides a shared starting context, persistent lessons, task state, structured inter-agent communication, review loops, vault-backed secret handling and an operator-facing control surface. Cloud and local models can work under the same operating conventions.
Shared context
Agents re-sync against the same mission, rules, project state and architecture before taking work.
Persistent operational memory
Approved lessons and decisions survive individual sessions and model changes.
Human control
Task boards, dashboards, alerts and approval gates keep operators in the loop.
Traceability
Work, commits, operational events and documentation can be tied back to system activity.
Source & documentation · Live demo · Robot Ross technical file
Research & open technical work
Graceful Degradation in Subscription-Constrained Multi-Agent Orchestration Systems
Operational study based on production telemetry from a multi-agent engineering fleet.
Reading Disposition from Smartphone-Class Open Models
A forward-only J-space probe evaluated on Apertus-4B, Ministral-3B and Qwen3-4B.
Robotics, smart energy and industrial systems
Peer-reviewed work includes microrobotics, smart-home energy systems and industrial interoperability.
Engineering engagements
Technical assessment
Review an AI, automation or industrial software system before committing to a major implementation.
- Architecture and integration review
- Model / local-vs-cloud assessment
- Operational and governance gaps
- Implementation plan and technical risks
AI & industrial systems integration
Hands-on work connecting models, software, machines, data sources and operator interfaces into a deployable system.
Agent-fleet deployment
Installation and configuration of Flotilla for engineering teams that want persistent multi-agent workflows, local-model support and an operator control plane.
Technical program support
Architecture, technical program management and engineering leadership for complex cross-disciplinary projects.
Talk to an engineer.
For project work, architecture reviews, Flotilla deployments or technical collaboration.