PRINCIPAL EMBEDDED SOFTWARE ENGINEER
I’m Jaydip Patel, a Principal Embedded Software Engineer specializing in Linux USB Host and Device drivers, silicon bring-up, and secure data streaming. I develop system-level software that connects hardware capabilities with real-world applications.
NOW MosChip · Contractor to Xilinx / AMD
01 / ABOUT
I work at the intersection of embedded platforms, operating systems, and the interfaces that connect them.
My experience spans C-based embedded development, Linux kernel drivers, and applications on Linux and Stratus VOS using C, PL/I, and COBOL. I bring a practical approach to hardware–software integration, production debugging, and root-cause analysis.
Since June 2023, I have worked as a Principal Engineer at MosChip, consulting for AMD / Xilinx. As an individual contributor, I take technical ownership of AI Engine issues across teams, develop Linux drivers, application drivers, and firmware, and mentor engineers and interns.
I actively use AI in my development work at AMD, with experience in Claude, Gemini, and ChatGPT. I also contribute to AI-assisted debugging tools and develop reusable MCP integrations and skills for team use.
Find me on LinkedIn ↗02 / SELECTED EXPERIENCE
On-site silicon bring-up, USB Linux drivers, and embedded platform engineering.
Read the case studies ↓
On-site assignment in San Jose, California, USA, for client Axiado Corporation while working as Senior Tech Lead at VVDN Technologies.
Hands-on silicon and board bring-up, alongside responsibilities for product requirements, software architecture, and engineering team leadership.
USB 2.0 / 3.0 Host controller (xHCI) and USB 2.0 Device Controller (UDC) driver development for Corigine USB IP. FPGA validation on HTG-960, with bitstream programming through Vivado and JTAG / TRACE-32.
Host and Gadget integration and validation on Tyan servers, plus virtual bridging for Mass Storage and HID device enumeration.
Designing and implementing Linux driver features, enhancing firmware, and resolving driver and platform issues. Leading the AI Engine device driver team through design, implementation, testing, and release.
Qualcomm SDK builds and Linux 5.4 platform integration for a smart speaker and access point. Device tree configuration, I2C/SPI/UART enablement, interface validation, and WiFi 6E over PCIe.
Dynamic library development to intercept openVOS file operations, with AES-128/256 CTR encryption, PL/I-to-C data integration, online re-encryption, and tooling to inspect encrypted files.
Porting and integrating librdkafka on Stratus VOS, with Kerberos authentication to support secure data transmission.
Linux board support packages, open-source library porting, and hardware–software integration across embedded platforms, alongside Tuya-based product and hardware development.
ENGINEERING / CASE STUDIES
Three stories of ownership, investigation, and making engineering knowledge useful to others. Select a story to explore.
At VVDN, I took ownership of USB Linux driver development for client Axiado as silicon bring-up approached. The vendor supplied a bare-metal implementation, and USB still needed Linux Host and Device controller support. I understood driver fundamentals, but this was my first opportunity to build this kind of controller driver from the ground up.
I used an available MediaTek development board to study a working Linux USB stack. I built kernel images with Yocto, reviewed the driver source, and added trace messages to understand the execution flow. That hands-on preparation gave me a foundation for understanding the vendor implementation and adapting it to Linux.
I developed Linux support for USB Host and Device controller operation from the supplied code. The work included USB 2.0 / 3.0 Host controller and USB 2.0 Device Controller development, FPGA validation, and an on-site silicon and board bring-up assignment with Axiado in San Jose, California.
I delivered the Linux Host and Device driver implementation and took the work into hands-on bring-up. The project strengthened my ability to learn an unfamiliar subsystem through source-code investigation and carry that understanding into hardware validation.
Internal and customer-reported JIRA issues can require investigation across several software and hardware layers. Much of the knowledge needed to narrow down an issue lives in engineers’ experience. As part of an AI initiative during my AMD engagement, I contributed to making that knowledge reusable through Debug Assist.
I developed MCP integrations and skills that captured my debugging methodology as structured playbooks. These helped AI agents analyze tickets, investigate likely causes, and suggest possible fixes. I built on the agentic AI infrastructure developed by colleagues, contributing the engineering skills and domain knowledge.
I raised the need to improve skills when AI-generated analysis missed important context. Working with the AI architecture team, I provided feedback on a workflow for capturing investigation findings and using them to create or refine reusable skills. I also developed skills for acquiring and booting shared development boards.
The debugging skills became available for use across teams, and repeated board setup steps could be invoked through an AI agent. My contribution turned practical investigation experience into tools that colleagues could reuse and improve.
An issue report described an inference failure with little information about its cause. The execution path crossed the application, runtime, application driver, Linux kernel, and hardware, making the source of the failure unclear.
I worked with the reporter to obtain reproduction steps and reproduced the failure on a development board. I examined kernel logs and hardware error events, then consulted the relevant documentation to understand what the failure indicated.
My initial suspicion about the failing component did not resolve the issue. I built and investigated the components along the execution path, including code outside my team’s direct ownership. By isolating read and write operations and varying the transfer size, I narrowed the failure to a hardware-access operation in the driver path.
I resolved the failure by changing the access pattern to respect the hardware’s supported transaction size. The investigation reinforced the value of a reproducible test case, controlled experiments, and revisiting an initial hypothesis when the evidence points elsewhere.
AI / ENGINEERING WORKFLOWS
Applying AI to everyday development, issue analysis, and shared lab workflows at AMD / Xilinx.
Contributed to Debug Assist, an internal AI tool for analyzing JIRA tickets covering internal and customer-reported issues. Developed debugging skills to help narrow down likely causes and guide investigation and potential fixes.
Built on the agentic AI platform developed by the wider team, contributing domain knowledge from embedded and systems debugging.
Developed MCP integrations and reusable skills that turn my debugging methodology into structured playbooks for AI agents. Made these available for engineers across teams to use in their own investigations.
Contributed engineering feedback to help refine the skills as new issues and gaps are discovered.
Developed reusable skills to acquire available development boards from a shared network lab and run the board boot workflow.
Captured repeated image preparation, flashing, and reboot steps in skills that the team can invoke through an AI agent when preparing hardware for development or issue reproduction.
03 / OPEN SOURCE
I contribute to public Xilinx repositories as part of my systems software work. This space will map those patches by repository, date, and area of contribution.
Public commits attributed to jaydipamd in official Xilinx and AMD repositories. Enable JavaScript to explore the snapshot.
PUBLIC COMMIT EXPLORER
CAREER / EXPERIENCE
Jun 2023 — Present
Consultant at AMD / Xilinx. Individual contributor with technical ownership of AI Engine Linux and application driver development, cross-functional issue investigation, and mentoring of engineers and interns. Active use of AI for development, contributions to Debug Assist for JIRA issue analysis, and development of reusable MCP integrations and skills for debugging and board setup.
Dec 2021 — Jun 2023
On-site silicon bring-up · San Jose, California, USAClient assignment for Axiado Corporation during my time at VVDN, covering silicon and board bring-up. Also responsible for product requirements, software architecture, and engineering team leadership.
Oct 2019 — Nov 2021
Freelance software services and maintenance consulting.
Jan 2015 — Sep 2019
Worked for Application Resource Inc., New York. Software design, implementation, QA, customer support, and remote production debugging.
Jan 2013 — Jan 2015
Telecom product features, device drivers, hardware test software, and production release validation.
EDUCATION
B.E. in Electronics and Communication · 2012
04 / TECHNICAL TOOLKIT
C · Embedded C · C++ · PL/I · COBOL · Basic Python · Linux · Stratus VOS
Linux BSP · Yocto · Buildroot · Kernel drivers · Tuya
USB Host (xHCI) / Device (UDC, Gadget) · I2C · SGPIO · SPI · QSPI · Ethernet · 10G fiber · TCP/IP sockets
ARM Cortex-A53 / A72 / M0 / M55 / R5 · Qualcomm IPQ8074A / IPQ6010 · MediaTek MT8365 · Xilinx AI Engine VCK190
Claude · Gemini · ChatGPT · AI-assisted development · Debug Assist · MCP integrations · Reusable agent skills · Board setup automation
Git · SVN · JIRA · Agile · Production debugging · Root-cause analysis
05 / WHAT’S NEXT
Hiring for systems expertise, exploring a project, or looking to collaborate? Let’s connect.
write2jaydeep@gmail.com · +91 90993 07746