Durable Task Runtime
Distributed execution and benchmarking infrastructure for reliable workload processing, controlled recovery, and performance investigation.
Read the record →Software infrastructure and embedded systems
Software that keeps its record when the system fails.
Herder Elektronische Systemen builds high-reliability software infrastructure for systems that execute work at scale, recover from failure, and need measurable performance. The firm also develops the embedded control and validation systems that connect software to physical equipment.
The work is grounded in durable state, distributed coordination, secure processing, hardware-backed firmware, and test systems that show what actually happened.
What the firm builds
Software platforms first. Embedded systems where they make the platform real.
Distributed execution systems
Coordinator and worker software that assigns work, retains durable state, recovers after failures, and measures the cost of every layer.
Primary capabilitySecure operational platformsSystems that control identity, preserve provenance, process untrusted input carefully, and retain review and recovery records.
Primary capabilityEmbedded and real-time systemsFirmware, device interfaces, control behavior, and target-side records for equipment that must agree with the software around it.
Engineering capabilityValidation and test infrastructureLinux targets, test orchestration, instrumentation, and data workflows that make results repeatable and useful.
Engineering capabilityFlagship engineering record
Distributed execution infrastructure built to be measured, changed, and recovered.
| The system | The Durable Task Runtime distributes deterministic workloads across workers while keeping durable task state and a fixed correctness baseline. |
|---|---|
| The failure case | Workers and coordinators can stop during execution. The system must account for completed work, reject stale results, requeue unfinished work, and recover without operator reconstruction. |
| The evidence | A 100,000-job chaos run included 17 worker crashes and 2 coordinator restarts with 0 unaccounted tasks. |
| Why it matters | It gives the firm a controlled platform for performance engineering: change one architectural variable, rerun the workload, and prove that an optimization did not compromise correctness. |
Engineering work
Evidence from software, embedded, and validation systems.
Enterprise processing workflow
Controlled identity, provenance, hardened archive handling, review steps, and recovery across a private multi-service product.
Read the record →Dual-core motion testbed
Target-side data capture, shared-memory coordination, and host tooling for repeatable motion characterization.
Read the record →Dual-PMSM dynamometer
Motor-control integration and real-time test workflows for measured drivetrain validation.
Read the record →Project inquiry
Bring the failure mode, workload, or product boundary.
Describe the system, what it must do reliably, what evidence exists, and what decision the work needs to support.
Discuss an engineering problem