Senior Applied Scientist, On-Device ML

About the role

The will spearhead the design and implementation of machine learning models specifically architected for direct execution on edge hardware within critical infrastructure environments. This position is the cornerstone of advancing Gridware's active grid response capabilities, which monitor electrical, physical, and environmental grid dynamics to significantly enhance reliability and safety outcomes. The successful hire will own the complex trade-off analysis required to balance model precision against the severe constraints of power consumption and memory availability in field-deployed sensors. You will translate abstract grid domain concepts into robust, efficient algorithms that operate reliably without constant cloud connectivity, embedding intelligence directly at the source of data generation. This role demands a unique fusion of applied research prowess, algorithmic innovation, and rigorous production engineering discipline to ensure models are not just accurate but deployable in resource-limited scenarios. You will collaborate intimately with hardware and firmware teams to ensure that the machine learning stack integrates seamlessly with drivers, sensors, and low-level system firmware. Ultimately, your work will directly determine the grid's capacity to detect anomalies early and execute proactive, reliable responses in real-time. For more information on our mission and technology, visit www.Gridware.io.

Key facts

What you'll do

• Architect and develop on-device machine learning models that function within the strict power and memory budgets of grid sensor hardware, ensuring operational viability in constrained environments.
• Implement low-level algorithms in Python and C++ that are optimized for edge deployment, rigorously adhering to standards for efficiency, safety, and long-term reliability in critical infrastructure.
• Measure and optimize the intricate trade-offs between model accuracy, inference speed, and energy consumption to determine the most effective deployment strategy for specific grid monitoring use cases.
• Work directly with firmware engineers to synchronize drivers, sensors, and machine learning models, ensuring seamless communication and data flow across the entire hardware-software stack.
• Translate complex grid domain signals, such as electrical and environmental dynamics, into reliable, stable features that are suitable for consistent inference on embedded platforms.
• Verify that on-device behavior remains consistent and dependable across a vast array of operating conditions, environmental variables, and potential data shifts that could impact model integrity.
• Partner with product teams to align technical model capabilities with the practical, real-world requirements of grid management and active grid response operations.
• Document methodologies, experimental results, and model behaviors comprehensively to empower operations and engineering teams to understand, maintain, and extend the system throughout its lifecycle.
• Conduct rigorous testing and validation protocols to ensure model resilience and performance in the real world, far beyond controlled laboratory conditions.
• Lead the evaluation and integration of on-device machine learning frameworks and model optimization methods to push the boundaries of what is possible with edge intelligence in critical infrastructure.

Requirements

• Bring hands-on experience with on-device machine learning frameworks and model optimization methods, demonstrating a deep understanding of the field.
• Show strong programming skills in Python and C++, which are essential for implementing efficient and reliable algorithms for edge hardware.
• Exhibit proficiency in numerical computing libraries used for processing sensor data, enabling effective analysis and transformation of complex grid signals.
• Possess a minimum of three years of experience building models specifically for embedded or edge platforms, with a proven track record in production environments.
• Demonstrate the ability to handle multimodal time-series inputs while strictly respecting defined latency and resource budgets that are critical for grid operations.
• Adept at navigating the constraints of embedded systems, where decisions must balance precision, energy efficiency, and computational limits without compromise.
• Capable of validating model performance under diverse and dynamic conditions to ensure safety and reliability are never compromised.
• Committed to maintaining the highest standards of code quality, documentation, and collaborative engineering practices essential for critical infrastructure technology.

Nice to have

No additional requirements or preferences are specified at this time.

Practical notes

-

Location: San Francisco, California

-

Engagement: Full-Time

-

Compensation: Base salary of $243,000, with a target total compensation of $332,000 for 2024.

• Please confirm all details on the official application page.