AI
Embedded Software Engineer II

ArianIzadi.

I build software close to the hardware: perception, control, low-level systems, and the tooling around them.

See work ↓Résumé ↗

Approach

Software thatholds up whenthings go wrong.

Perception pipelines with real timing and memory budgets. A bare-metal kernel from boot up. Networked control that fails safe when the link drops.

Failure handling is designed up front, not bolted on.

01Selected work

Projects

Seven projects. For each: the problem, what I built, and what came of it.

  1. 100k pts / frameC++ · PCL10 Hz
    01Koshee AI
    01Koshee AI · Robotics Perception

    LiDAR Perception Pipeline

    Problem
    High-density point-cloud processing had to stay within strict timing and memory constraints.
    What I built
    Created a C++/PCL LiDAR perception pipeline that handled more than 100,000 points per frame.
    Result
    Processed high-density LiDAR data within explicit timing and memory constraints.
    • C++
    • PCL
    • LiDAR
    • Point Clouds
    Work project, source not public
  2. ControlRobotUDP heartbeat< 2.0 shalt
    02Koshee AI
    02Koshee AI · Fault Handling

    UDP Heartbeat Safety Switch

    Problem
    Loss of the network control link required a bounded response.
    What I built
    Used UDP heartbeats to detect network partitions and connect communication loss to shutdown control.
    Result
    Enforced hard shutdown behavior within a sub-two-second window.
    • UDP
    • Networking
    • Heartbeat Monitoring
    • Fault Handling
    Work project, source not public
  3. RISC-V> boot ok> uart 115200> hello, worldboot.S → linker.ld → kernel.rsQEMU
    03Low-Level Systems
    03Low-Level Systems

    RustOS: Bare-Metal RISC-V Kernel

    Problem
    Understanding operating-system internals requires direct control over boot flow, memory layout, device I/O, and toolchain constraints.
    What I built
    Implemented a boot path in RISC-V assembly, custom linker scripts, UART output, and a small Rust kernel environment.
    Result
    Exercised low-level debugging across assembly, Rust, QEMU, and bare-metal constraints.
    • Rust
    • Assembly
    • RISC-V
    • QEMU
    • Linker Scripts
    Source ↗
  4. mainf0f1f2f3f4f5LLVM IR → call graph → cyclescritical pathloop
    04Heuristic Timing Analysis
    04Heuristic Timing Analysis

    T-REX: WCET Analysis for Rust

    Problem
    Timing analysis for Rust code needs a bridge from LLVM IR into call graphs and loop/cycle analysis.
    What I built
    Parsed LLVM IR, built call graphs, detected cycles, and visualized timing paths with Python tooling.
    Result
    Explored heuristic WCET workflows for systems where predictable execution matters.
    • Python
    • LLVM
    • Rust
    • NetworkX
    • Matplotlib
    Source ↗
  5. plaintextSEAL ctxciphertextC++17 · CMake · GTest · CIkeys
    05C++17 Library
    05C++17 Library · Testing

    SEALCrypt

    Problem
    Microsoft SEAL setup has enough context and parameter wiring to slow down experimentation with encrypted computation.
    What I built
    Wrapped encryption-context initialization behind a modern C++17 interface with CMake, tests, and CI.
    Result
    Reduced setup overhead while keeping configuration and testability explicit in the library interface.
    • C++17
    • CMake
    • Google Test
    • GitHub Actions
    Source ↗
  6. SwiftUI · NetworkExtensionMasterDnsVPNDNS :53go · gomobile bridge
    06iOS Packet Tunnel
    06iOS Packet Tunnel

    VpnDad

    Problem
    MasterDnsVPN is powerful on the server side, but it did not have a practical iOS client for family members who just need a reliable connect button.
    What I built
    Built a SwiftUI app with a Network Extension packet tunnel, a Go mobile bridge, JSON profile import/export, diagnostics, and MasterDnsVPN-aware reliability controls.
    Result
    Turned a command-line DNS tunnel into a mobile workflow that can run against standard MasterDnsVPN servers while exposing custom profile controls when the server supports them.
    • SwiftUI
    • NetworkExtension
    • Go
    • gomobile
    • MasterDnsVPN
    Source ↗
  7. class: railIoU 0.91PyTorch · Lightning
    07Semantic Segmentation Framework
    07Semantic Segmentation Framework

    Segmentary

    Problem
    Semantic segmentation experiments are difficult to reproduce when configs, taxonomy mappings, checkpoints, metrics, and deployment benchmarks live in separate tools.
    What I built
    Built a config-driven PyTorch suite for training, transfer learning, evaluation, comparison, export, and analysis across Hugging Face, native, and SMP models.
    Result
    Kept training, evaluation, export, and benchmark evidence together so experiments remain reproducible and comparable.
    • Python
    • PyTorch
    • Hugging Face
    • Lightning
    • CUDA
    Source ↗
02Experience

Where I'veworked

Embedded R&D, banking backends, robotics, and teaching.

  1. 0101

    July 2026 to Present

    Konami Gaming, Inc.

    Embedded Software Engineer II

    Las Vegas, NV

    R&D work across embedded hardware, manufacturing systems, and regulated gaming software.

    • Research, design, and maintain C/C++ components integrated with embedded hardware and manufacturing systems.
    • Translate complex requirements into testable, standards-aligned software through code review, unit testing, revision control, and clear documentation.
    • Coordinate delivery across Engineering, QA, Training, and Documentation while providing technical guidance to junior developers.
    • C/C++
    • Embedded Systems
    • Hardware Integration
    • Unit Testing
    • Technical Leadership
  2. 0202

    July 2025 to July 2026

    Credit One Bank

    Software Engineer

    Las Vegas, NV

    Backend engineer working on event-driven account and customer systems in a regulated banking environment.

    • Develop and maintain Java/Spring Boot microservices that process account and customer events through Kafka-based workflows.
    • Implement Kafka consumers and producers with tested error handling, retry behavior, and dead-letter queue flows.
    • Support Elasticsearch and relational database integrations for account search, indexing, and event-driven synchronization.
    • Java
    • Spring Boot
    • Kafka
    • Kubernetes
    • Elasticsearch
  3. 0303

    Jan 2024 to July 2025

    Koshee AI

    Embedded Systems & Robotics Intern

    Built robotics systems across communication-failure handling, LiDAR perception, and C++ build automation.

    • Architected a UDP heartbeat shutdown control that detected network partitions and enforced hard shutdown behavior within a sub-two-second window.
    • Created a LiDAR perception pipeline in C++/PCL that handled 100k+ points per frame within strict timing and memory constraints.
    • Built CMake and GitHub Actions workflows with distributed caching, reducing build times from 15 minutes to 4 minutes.
    • C++
    • CMake/CTest
    • LiDAR
    • Networking
    • GitHub Actions
  4. 0404

    May 2023 to Jan 2024

    Code Central

    Lead Instructor & Full Stack Web Developer

    Taught programming while improving an internal learning platform used by students and schools.

    • Improved LMS data protection by patching SQL injection issues and strengthening PHP/MySQL data access patterns.
    • Implemented Clever SSO so 200+ students could access course material through school accounts.
    • Led classroom instruction across web fundamentals, backend concepts, and secure programming habits.
    • PHP
    • MySQL
    • JavaScript
    • API Development
    • Secure Programming

03Skills

What I
work with.

Most of it comes down to memory, timing, interfaces, and being able to prove the thing works.

01Embedded & SystemsProduction software, operating-system tooling, timing constraints, and networked fault handling.
  • C/C++
  • Linux
  • Embedded Systems
  • Real-Time Systems
  • Multithreaded Systems
  • Networking
02Low-Level / HardwareBare-metal boot flow, architecture, device output, and hardware/software boundaries.
  • Rust
  • RISC-V
  • QEMU
  • UART
  • Assembly
  • GDB
  • Hardware Integration
03LanguagesLanguages used across embedded, systems, research, and production service work.
  • C
  • C++
  • Rust
  • Python
  • Java
04Tooling & DebuggingBuild, test, inspect, and version software across low-level and production environments.
  • CMake/CTest
  • Google Test
  • Git
  • GitHub Actions
  • LLVM
  • Ghidra
  • IDA Pro
05AdditionalRobotics perception, computer vision, backend systems, and technical interfaces.
  • Robotics
  • LiDAR
  • PCL
  • OpenCV
  • PyTorch
  • Kafka
  • Spring Boot
  • TypeScript
  • Next.js
04More projects

Everythingelse

Smaller tools, experiments, and side projects.

  1. 01LiDAR 2D Room MappingRobotics PerceptionProcessed point clouds with PCL and OpenCV, removed floor planes with RANSAC, and generated 2D room maps.C++ / PCL / OpenCV / CMake / LiDAR
  2. 02Reverse Engineering TutorialsLow-Level SecurityDocumented beginner crackme and binary-patching walkthroughs using Hopper and Ghidra.C++ / Python / Assembly / Ghidra / Hopper
  3. 03Image / Video SuiteHardware-Aware ML ToolingBuilt a modular Python CLI with model adapters, hardware and VRAM inspection, process detection, GPU recommendations, and soft resource locks.Python / CUDA / Hugging Face / Diffusers / Rich
  4. 04Inference CheckerModel Evaluation ToolingBuilt an interactive Next.js canvas tool with single, side-by-side, diff, overlay, hover inspection, and per-class metric modes.Next.js / TypeScript / Canvas API / Tailwind CSS
  5. 05CalPrivacy-First iOS AppBuilt a local-first SwiftUI app that reads package labels with on-device Vision OCR and stores nutrition data locally.Swift / SwiftUI / SwiftData / Vision / HealthKit
  6. 06Fantasy Football NotifierReal-Time AlertingBuilt a Python pipeline that joins X and RotoWire reports with ESPN and Sleeper rosters, deterministic depth charts, model-assisted classification, deduplication, and Telegram delivery.Python / OpenRouter / Telegram / X API / Sleeper
  7. 07Knowledge MapperAI Learning ToolBuilt a Next.js assessment interface that generates dynamic checks and maps knowledge gaps.Next.js / TypeScript / React / LLMs
  8. 08Lazy WordlerWeb UtilityBuilt a small Next.js utility for answer lookup and solver experimentation.Next.js / TypeScript / React / Tailwind CSS
  9. 09Proton Pass to Chrome ConverterMigration UtilityBuilt a simple Python converter from Proton Pass export shape into Chrome-compatible CSV output.Python / JSON / CSV

Open source / external

Open source

  1. OctoMap / PCD File ReadingAdded PCD point-cloud file reading support to a widely used 3D occupancy mapping library.Merged
  2. MasterDnsVPN / iOS Mobile Bridge and Reliability ControlsExtended a DNS-tunneled proxy stack with a gomobile-friendly bridge, app diagnostics, profile-driven settings, and negotiated reliability experiments for mobile use.Fork
  3. MMSegmentation / RailSem19 Transfer LearningMaintained railway segmentation training changes for model experimentation and research.Fork

Education / UNLV

Education.

01

M.S. Computer Science

4.0 GPA

  • Real-time & embedded systems
  • Advanced operating systems
  • Network security
  • Railway track segmentation research

02

B.S. Computer Science

Jul 2024

  • CyberFire CTF: 1st place
  • NCL CTF: top 8% nationally
  • Bosch Future Mobility Challenge
  • President of Layer Zero

Certification / AWS Certified Cloud Practitioner