The need
Understand vehicle behavior from large CAN traces, compare signals and find events across many logs.
ENGINEERING TOOLS / AUTOMOTIVE / DESKTOP
From raw logs
to a clearer picture.
I build applications that connect to hardware, decode logs and help engineers investigate what happened. My experience comes from six years in automotive software, working with CAN, serial communication and data analysis.
SELECTED PROJECT
RoadMetrics
A Windows desktop application for vehicle log investigation
My role: built the complete system, including the desktop application, analysis engine and AI integrations.
What I built
BLF, ASC, MF4 and TRC decoding with DBC and metadata. Numeric and enum signal plots, a rules engine and parallel batch analysis.
What it enables
Jump from a detected event to its time span on a plot, compare logs, bookmark findings and use calculated parameters in the same investigation tool.
C# / .NETWPFPython / FastAPICAN / DBCAzure OpenAIAzure AI SearchLangGraphRedisMCP
Desktop and hardware
Experience building C#/.NET/WPF applications and connecting desktop tools to embedded systems through CAN and serial communication.
Analysis and anomaly detection
Thresholds, ranges, event duration, rate of change, stalled signals and enum transitions. Batch analysis and lazy decoding for large files.
Context-aware investigation
CLI and MCP expose the analysis engine to tools such as Cursor and ChatGPT, without requiring the desktop UI.
AI + ENGINEERING CONTEXT
Ask about your logs.
Check the answers.
In RoadMetrics, the assistant drafts detection rules from natural language using the parameter catalog, with apply and undo controls.
I also connected a Cursor environment to log analysis, code and company data for contextual investigation. The goal is to make analysis and insight extraction easier, while keeping findings verifiable against the source.
My focus is integrating existing AI models into products and workflows, rather than training foundation models.