IDES, Inc.

IDES is a JEOL group subsidiary working in the transmission electron microscope (TEM) market. IDES develops Relativity sub-framing system, electrostatic dose modulation system, and various other TEM peripheral systems to support scientific applications.

As the primary firmware engineer at IDES, I focused on maintaining and extending existing firmware, developing new voltage control firmware for our flagship product, and performing hardware bringups. Our firmware ecosystem is mixed C/C++ across Xilinx Zynq SoC and Atmel/Microchip platforms. Additionally, I prepared embedded Linux system images for production of an edge computer vision platform.

Working at a small company, I also took on responsibilities outside of my core firmware role including electronics manufacturing, ECAD support, PCBA procurement, and developing user-facing Python software for hardware interfacing.

IDES recognized my contributions and I was quickly promoted to senior engineer.

As senior engineer, I continued supporting existing hardware/firmware shipped to customers, including working with our parent company to diagnose and test systems internationally. I also took on mentoring responsibilities, assisting in hiring and training of manufacturing and software engineers.

My major firmware project was transitioning our Xilinx codebase to support multiple SoC variants while maintaining a unified codebase. This involved building out a development process for multiple Vivado projects with shared Verilog HDL and a unified C/C++ system supporting multi-core Vitis projects.

I developed a SCPI-based voltage instrument control service and implemented an AXI DMA transfer pipeline that improved memory throughput by 10x for a key application.

I also led the transition from a bare-metal round-robin architecture to FreeRTOS, utilizing dual cores to enable onboard application-specific acceleration. This included spinning up a REST API and embedded web dashboard built on FreeRTOS+TCP.

I trained manufacturing engineers in electronics assembly processes and provided ongoing QA support. I authored extensive build documentation, standardized packing procedures, and maintained our MRP system. Over time I was able to transition out of active time on this role to be minorly supporting.

I led component selection for next-generation products, evaluating microcontrollers, ethernet controllers, and I2C devices with a focus on maintainability and supply-chain resilience.

I designed and developed MiniConsole, a USB user-interface peripheral featuring integrated OLED displays, a tactile encoder, and haptic/audio feedback. I took this product from requirements analysis through schematic capture in Altium, board bringup, firmware development, mechanical enclosure design, custom Windows driver development, Python interface development, integration into the existing software ecosystem, CE marking and productization.

I led a major refactor of our hardware control backend, replacing a tightly coupled architecture with a decoupled PubSub-based system (drawing on my experience with ROS). I built out a layered testing strategy spanning driver-level unit/integration tests through system-level QoS validation. I also mentored an intern who prototyped the new REST API layer.

My current focus is building a real-time vision system on the Nvidia Holoscan SDK for 4D-STEM applications. The system processes multiple concurrent image streams from high-speed detectors and generates signal control feedback to the microscope, keeping acquisition and instrument response in a tight loop. This work spans GPU pipeline development, multi-channel stream handling, and integration with our existing hardware control ecosystem.

My role has expanded to include AI-augmented development workflows for firmware and software tasks. I work daily with agentic coding tools, both Claude and Gemini CLI, choosing between them based on their aptitude for the task at hand. These tools assist with debugging, code generation, and maintenance across our embedded and application-level codebases.

To bring these tools into our engineering process safely, I developed an agentic-first, MCP-based CI system for the Holoscan vision project. It gives coding agents a controlled interface to trigger builds, run benchmarks, and deploy against microscope hardware resources without putting the instrument at risk.