Experience

Showcasing all projects

Google2021 - Present

Camera ISP Register Generation Toolchain

Software Engineer - L5
Mountain View, CA
  • Spearheaded a complete overhaul of the ISP register generation codebase, reducing SOC engineering hours from months to days via automated IP-XACT parsing.
  • Orchestrated a year-long cross-functional migration roadmap, managing stakeholder expectations and ensuring 100% on-time delivery for critical SOC milestones.
  • Engineered a full-stack register visualization and diagnostic suite, automating the identification of duplicate registers and reducing the time it takes to debug register issues from hours to minutes.
  • Developed a high-performance C++ code generator that outputs packed memory structures and automated default-value population methods, eliminating manual programming errors and automating a process that takes months of engineering hours.
Google2021 – Present

Secure Camera Pipeline Lead

Software Engineer - L5
Mountain View, CA
  • Optimized the Camera HAL secure pipeline, reducing the permanent secure memory heap carveout by 50% through aggressive resource management.
  • Reduced end-to-end Face Authentication latency by 10% by optimizing ISP hardware utilization and streamlining cross-processor communication.
  • Lead cross-functional bring-up of the secure camera stack for annual SOC cycles, collaborating with Face Auth, Secure Realm (TEE), and Image Quality (IQ) teams.
  • Sustained 99.9% reliability for the device’s most frequent camera use-case by resolving complex race conditions and secure-buffer synchronization issues.
Google (Pixel Camera)20XX – Present

Kernel Abstraction Layer Lead

Software Engineer
Mountain View, CA
  • Direct the ownership and evolution of the Pixel Camera HAL’s Linux Kernel Abstraction Layer, overseeing the software-to-hardware interface for high-performance camera systems.
  • Optimized system resource utilization by 66%, consolidating 15 dedicated threads into a unified shared event system capable of multiplexing hardware interrupts.
  • Reduced IOCTL syscall overhead by 66%–80% by architecting a logical expression interface (AND/OR) for atomic hardware trigger programming.
  • Designed a multi-status Kernel Fence mechanism (OK/ERROR) to replace legacy binary signals, enabling robust error handling and complex transaction chaining.
Thermo Fisher Scientific2018 – 2021

Focused Ion Beam (FIB) Control Software

Software Engineer
Fremont, CA
  • Redesigned the real-time rendering engine for Focused Ion Beam (FIB) circuitry editing, increasing visual resolution accuracy to 99.9% and doubling the frame rate.
  • Optimized image processing and post-processing efficiency by over 100% through a comprehensive architectural overhaul of the C++/C# visualization pipeline.
  • Developed a full-stack CI/CD dashboard (Node.js/Angular), reducing manual build overhead by 3 hours per week and eliminating "tribal knowledge" bottlenecks.
Community Pool Club2020 – 2021

COVID-19 Reservation System for Community Pool

Full-Stack Developer (Volunteer)
Remote / Local
  • Engineered and deployed a full-stack reservation system to enforce COVID-19 safety compliance, supporting over 700 weekly reservations for 40+ families.
  • Developed a custom Node.js backend integrated with Google Sheets API, enabling non-technical staff to manage the database via a familiar spreadsheet interface.
  • Hosted the entire production environment on a Raspberry Pi, providing a low-cost, high-uptime solution for community members during government mandates.
  • Rapidly prototyped and launched the MVP within weeks to meet urgent regulatory deadlines, ensuring the facility could remain open to the public.
Personal Project2020 – Present

Cauldron: Custom IoT Halloween Special Effects System

Creator / Hobbyist Engineer
Home Lab
  • Engineered "Cauldron," a low-latency IoT system using a WiFi-enabled microcontroller and a Python-based control server to synchronize audio-visual effects.
  • Developed a real-time signal processing engine that dynamically modulates LED brightness and patterns based on decibel levels and audio frequencies.
  • Implemented a UDP-based communication protocol for minimal latency between the server (Python) and the hardware (Microcontroller), ensuring frame-perfect light-to-audio sync.
  • Integrated a real-time voice modulation and streaming pipeline, allowing for live, low-latency audio capture and demon/witch vocal effects.
© 2026 Tommy Kardach. Built with Angular & Tailwind.