Arseni Aliakseichyk
Embedded and Infrastructure Engineer
View Portfolio Photos, videos & technical deep-divesProfessional Summary
Embedded and infrastructure engineer, working from C firmware on microcontrollers to the Linux, cloud and CI/CD infrastructure around it. Currently a Software Engineer at GlobalLogic Poland: built from scratch a cloud data-ingestion module (Python, Kubernetes, Terraform, AWS) for a client’s AI tooling. In the 2026 internship there, built a robotics platform for three teams: C firmware on ESP32-S3 (FreeRTOS), a MicroPython API and self-hosted GitLab CI/CD. Also embedded Linux and edge ML on Raspberry Pi 5 (PyTorch → ONNX, Hailo AI HAT), a homelab with 13+ Docker services running 2+ years, and CyberGlove, my IoT smart glove on ESP32-S3.
Technical Skills
Experience
GlobalLogic Poland
Jun 2025 – PresentSoftware Engineer
Aug 2026 – Present- Designed and built from scratch, as the only engineer, a data-ingestion module for a client’s internal AI tools: it pulls support tickets, removes credentials and personal data, filters out irrelevant tickets with an LLM, and stores a clean, versioned corpus.
- Built the pipeline as a scheduled Kubernetes job (Python, Helm) on S3 and DynamoDB, with source adapters so a new data source plugs in without changing anything downstream; re-running it never duplicates or loses data.
- Wrote a redaction step that blocks credentials and personal data before anything is stored (zero misses and zero false positives on a hand-labeled sample), and an LLM relevance filter (Claude Haiku 4.5 on Amazon Bedrock) that only ever sees redacted text.
- Provisioned the AWS infrastructure in Terraform (S3, DynamoDB, KMS, IAM, an alert on a missed scheduled run) and packaged the job as a hardened container: non-root, read-only filesystem, dropped capabilities.
- Restructured the GitLab CI pipeline into modular files and built an offline end-to-end test environment (k3s in Docker). Earlier in the same module: a live email-capture pipeline (SES, S3, SQS, Lambda).
Embedded & Infrastructure Engineer
Jun 2026 – Sep 2026- Designed and built from scratch a robotics platform for three teams, who wrote the maze-solving algorithm in Python on top of it; owned the firmware, the platform and the infrastructure underneath.
- Wrote the drive firmware in C on ESP32-S3 (ESP-IDF, FreeRTOS, ~9,000 lines): a 200 Hz control loop with safety stops, mecanum drive, gyro heading hold and six ToF distance sensors, plus signed OTA updates with A/B slots.
- Designed a two-board setup: a MicroPython board gives the teams a simple, safe Python API over an authenticated UART link, with push-to-robot delivery of signed bundles and automatic rollback after crashes.
- Set up the infrastructure: self-hosted GitLab, an Ubuntu build server with per-team isolation, and CI for the firmware.
- Moved the platform from ESP32-C3 to ESP32-S3 after measuring its memory limits (the solver now runs on the robot at 19.4 Hz) and, from bench measurements, cut a 1.5 s start delay to 0.4 s.
- Wrote a 31-page API guide with examples tested against the real platform and a simulator for the teams; supported them in test sessions and presented sprint reviews (Scrum, Jira).
Embedded & IoT Developer
Jun 2025 – Oct 2025- Built the core embedded software for a modular robotic platform on Raspberry Pi 5: a C++17 multithreaded system with 5 concurrent threads handling motor PWM, sensor polling, MQTT, the GStreamer video pipeline, and a REST API server.
- Implemented real-time FPV video streaming via GStreamer (MJPEG over UDP, up to 1080p@60fps) and an HC-SR04 ultrasonic obstacle detection module in C with a 20 cm threshold and hysteresis.
- Stood up the deployment infrastructure: containerized the full stack with Docker Compose (Mosquitto MQTT broker + Redis) on a self-managed Linux server, automated with Bash scripts.
- Supported the team across the rest of the platform, including integrating the Hailo AI neural processor for human skeleton detection and obstacle avoidance - within an Agile/SCRUM team (Jira).
↳ Personal continuation after internship: Robot Platform - Consumer Firmware
Software & Infrastructure Engineer (Freelance)
Oct 2025 – Feb 2026Confidential B2B Project — Self-Hosted Linux Infrastructure & System-Level Android
Designed and deployed the project's self-hosted Linux backend: a Python notification worker in Docker on a private VPS, processing 10 event types with locale-aware delivery, exponential retry (5 attempts), and Firebase Cloud Messaging (FCM) token management. Built and maintained the full GitLab CI/CD pipeline (analyze → test → build → release) with artifact publishing. On the client side, wrote the native Android monitoring module in Kotlin (Device Owner API, Foreground Service, OOM adj 200).
Education
Pomeranian University (Uniwersytet Pomorski w Słupsku)
Oct 2023 – Feb 2027 (Expected)B.Eng. Computer Science - Programming in Engineering • GPA: 4.5 / 5.0
Relevant Coursework: Embedded Systems, Computer Architecture, Operating Systems, Software Engineering, Algorithms & Data Structures, Databases
Projects
CyberGlove - IoT Smart Glove Platform
Personal • 2025 – PresentMy flagship project - an IoT smart glove for real-time gesture control, in active development. Phase 1 (complete): ESP32-S3 firmware (C, ESP-IDF, FreeRTOS) with an ICM-20948 9-DoF IMU over SPI, custom fiber optic bend sensor via ADC, and 50 Hz WebSocket telemetry; hardware drafted in KiCad 8 on hand-soldered perfboard with a LiPo power chain. Phase 2 (in progress): five-finger bend sensing with an OpenSCAD 3D-printed optical fixture and on-device TinyML.
Robot Platform - Consumer Firmware
Personal • 2025Post-internship rewrite of the GlobalLogic robot's firmware in C11 as a single multithreaded binary (drivers + web server + networking). Zero-config UX: power on, join the robot's Wi-Fi hotspot, open the web panel - WebRTC video, joystick UI, MQTT control API, and an ESP-NOW low-latency channel via an ESP32-WROOM receiver. Demonstrated at a university open day. Planned next: a minimal custom Linux 6.x build.
Homelab Infrastructure - Edge Router + Production VPS
Personal • 2024 – PresentMy self-hosted network, end to end: an OpenWrt edge router (Wi-Fi 6) and a public VPS joined into one private network over a WireGuard site-to-site tunnel. Network-wide AdGuard DNS filtering, nftables default-deny segmentation (LAN / guest / WireGuard), system Nginx reverse proxy with Certbot TLS, and Authelia SSO + YubiKey 2FA in front of 13+ Docker services (2+ years uptime). Sanitized config, CI linting, and audit/health scripts published on GitHub.
STM32F4 Peripheral Firmware Development
Personal • 2024–2025Firmware on STM32F411 (main working board) and STM32F407VGT6 (ARM Cortex-M4 Discovery board introduced during GlobalLogic training) using the Low-Level (LL) driver library + CMSIS for register-level control. STM32CubeMX used only for initial clock tree configuration; peripherals (GPIO, timer-based PWM, NVIC interrupts, SPI/I2C/UART, ADC/DAC) driven entirely through LL from application code. Debugging with LA1010 logic analyzer and oscilloscope via JTAG/GDB.
Real-Time Shape & Symbol Classifier
Academic • 2025Full ML pipeline: PyTorch training (MobileNetV3, 80 classes) → ONNX export → edge inference on Raspberry Pi 5 at 25–30 FPS (640×480). Custom SPI LCD driver (ST7735S, RGB565) and HTTP dashboard for live classification stats.
ESP32 IoT Embedded Projects
Academic • 2024–2025Three projects on ESP32 (ESP-IDF/FreeRTOS): BLE gamepad monitor with NimBLE and SPI LCD, Firebase REST message display, and NTP clock with deep sleep power management.
UPSl - University Schedule Platform
Personal • 2024 – PresentFull-stack web platform actively used by students: FastAPI + SQLAlchemy backend, PostgreSQL, JWT auth with role-based access, and an admin CRUD panel. Runs in Docker Compose (PostgreSQL + FastAPI + Nginx) with Let's Encrypt SSL.
Languages
Side Projects
AstraBot - Discord music & management bot (Python, discord.py, yt-dlp/FFmpeg), deployed in production with ticket system.
Mess-Privat - End-to-end encrypted messaging over raw TCP sockets (Python, NaCl/libsodium, Curve25519, curses TUI).