Senior Software & Automation Engineer
Field instrument to cloud dashboard — I build the whole path.

I have worked across industrial operations, automation and software engineering since 2013. Most automation engineers stop at the historian and most software engineers have never commissioned a panel; I do both, usually on the same project, which is how you end up designing systems where the OT and IT halves actually fit together.

Day to day that means cloud architecture on AWS, Azure and GCP, backend and infrastructure work, and AI/ML for decision support and predictive maintenance — sitting on top of SCADA/HMI development, PLC and PID configuration, industrial protocols and field commissioning. Master's in Automation and Robotics (UFF), AVEVA-certified on Application Server.

The part I enjoy most is usually the constrained end: bare-metal C on STM32, RTOSes written from scratch, protocols implemented straight off the RFC, and navigation software that has to keep working when the satellite link drops.

• remote_play — an RFC 5389 STUN client written from scratch in C++ over Boost.Asio (binding request/response, XOR-MAPPED-ADDRESS parsing, transaction-ID validation), plus UDP hole punching and a hand-rolled reliability and session layer over UDP, driven from a PyQt6 GUI.
• line_follower_stm — competition line-follower robot: bare-metal C on STM32 with the LL library and a custom HAL, PID and Pure Pursuit control, virtual line following from pre-mapped track data, a custom PCB and a Fusion 360 chassis.
• line_follower_app_stm — its companion PyQt6 desktop app: telemetry, tuning and track mapping over Bluetooth.
• channel_state_information — the research pipeline behind the IEEE Access paper below: raw PCAP parsing of 802.11ac CSI from Broadcom chipsets, PyTorch training with mixed precision on CUDA, and an 18-model position-specific ensemble.
• Bosch_XDK110_Data_collection — sensor configuration and data collection for the IIoT instrumentation behind the IoT Forum paper.
• predictive_maintenance_on_azure_dataset — an end-to-end predictive-maintenance study on the Azure PdM dataset: extraction, preprocessing, clustering, classification and regression.

• Quality Control and Fault Detection with Statistics and Machine Learning: A Data-Driven Business Model — first author, 2023 IoT Forum. IIoT anomaly detection on an instrumented compressor motor: KNN with Dynamic Time Warping for classification and LSTM for failure forecasting, projecting a 4.23-point cut in failure time, from 6.45% to 2.22%.
• Subcarrier Selection for HAR Using CSI and CNN: Reducing Complexity and Enhancing Accuracy — IEEE Access (2024), presented at IEEE VCC 2024. Statistical subcarrier selection feeding a Conv1D CNN: over 98% classification accuracy on 20–40 subcarriers instead of 60+.