Senior AI/ML Scientist building signal-driven diagnostics, uncertainty-aware models, and LLM-powered agents for real-world industrial systems.
currently
Profile
I am a senior AI and machine learning scientist with a PhD in Engineering from NTNU and experience developing deployment-focused analytics for industrial assets. I work across signal processing, acoustic analytics, predictive maintenance, anomaly detection, uncertainty quantification, model validation, and production-oriented dashboards.
My applied work spans high-frequency operational sensor data, vibration-based condition monitoring, acoustic signal analysis, control system signals, IoT workflows, and structural health monitoring — across sectors including rail, energy, manufacturing, and infrastructure.
I build practical machine learning and signal analytics workflows that turn operational sensor data into validated diagnostics, calibrated model insight, and decision support for real-world maintenance and monitoring. Increasingly this includes LLM-powered agents and automation tooling that connect AI assistants directly to operational data and analysis pipelines, always with human review and deterministic logic where the engineering judgment matters.
Based in Trondheim, Norway, I work at the point where signal processing, machine learning, and real engineering constraints meet — building systems that engineers actually trust and use.
Organizations I've worked with
Stats from Google Scholar
Current focus
Selected work
2025 - Present
Physics-informed signal processing and ML pipelines for machinery health, combining sensing, feature extraction, and anomaly detection into structured diagnostic reports. Consolidated fragmented, multi-signal analysis into a single reproducible workflow that turns raw sensor streams into decisions an engineer can act on.
2025 - Present
A human-in-the-loop agent (Claude + Model Context Protocol) that orchestrates tool-use and retrieval over operational sensor data, while deterministic Python handles the inference and a human reviews every flagged case. Consolidated multi-signal condition-monitoring analysis into a single agent-driven workflow.
2024 - Present
Analytical pipelines linking continuous measurement signals to degradation behavior, applied to rotating machinery and rail wheel wear. Surfaced driver- and operation-related factors influencing wear that raw electrical signals alone missed, feeding maintenance and control decisions.
Research
Open to senior AI/ML roles, consulting, and collaborations where signal data, uncertainty-aware evaluation, LLM-powered automation, and reliable industrial decision support matter.
Get in touch