Skip to content
View syahrilTGR's full-sized avatar
3 followers · 3 following

Highlights

  • Pro

Block or report syahrilTGR

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
syahrilTGR/README.md

Hi, I'm Muhammad Syahril Eka Pratama 👋

Edge AI & Embedded Systems Builder | Real-Time Telemetry & Systems Optimization

Undergraduate student in Digital Telecommunication Networks (International Class) at Politeknik Negeri Malang. I engineer systems where resource-constrained physical hardware meets applied machine learning—focusing on deterministic firmware, edge vision inference, and compressed telemetry pipelines.

Portfolio • LinkedIn • Email

🛠️ Featured Engineering Builds

Note on Code Availability: Core repositories for production/thesis projects are maintained under private access due to commercialization NDAs and academic embargoes. Architecture specifications, test logs, and live code walkthroughs are available upon request during technical evaluations.

1. ESP32 Smart Quick Shifter & Ignition Controller

Status: Private (Commercial NDA with Automotive Partner)

  • The System: Designed firmware, circuit protection, and a custom Hardware-in-the-Loop (HIL) test harness for single-cylinder engine ignition management (Quick Shifter, Launch Control, Pit Limiter).
  • Engineering Highlights:
    • Built a deterministic HIL testbed: Python-based crank physics simulator (engine_sim.py) coupled with an ESP32-CAM 1 MHz hardware pulse generator (timerBegin).
    • Hardened ISR routines using IRAM_ATTR and FreeRTOS critical spinlocks (portENTER_CRITICAL_ISR), removing microsecond timing jitter.
    • Mitigated severe EMI and switch bouncing by combining H11L1 Schmitt-trigger optoisolators with a 4 ms software blanking window (interval >= 4000).
    • Shunted pulser signals via AQY212 PhotoMOS + 220Ω resistor to maintain ~42% crank pulse sync during cuts.
    • Achieved zero false-triggers across 3,500+ HIL validation pulses, locking a deterministic 84.8% cut duty cycle at >3,500 RPM with smooth 50 RPM hysteresis recovery.
  • Stack: C++, PlatformIO, ESP32, FreeRTOS, Hardware-in-the-Loop Testing, Python.

2. Eco-Bin: Continual-Learning Edge Waste Classifier

Status: Private (Undergraduate Thesis Research)

  • The System: A distributed computer vision pipeline with class-incremental learning and ultra-compressed over-the-air (OTA) state updates.
  • Engineering Highlights:
    • Decoupled architecture: Frozen YOLOv8n detector (13.8 ms/frame inference on MX150 GPU, mAP50 0.992) + MobileNetV3-Large feature extractor with an IPCHead (0 trainable parameters).
    • Engineered an asynchronous MQTT OTA synchronization pipeline using paho-mqtt over Mosquitto (QoS 1).
    • Compressed model update payloads from 16.26 MB down to 42.93 KB (387.8x bandwidth reduction) via Gaussian KDE prototype state serialization.
    • Zero-downtime ONNX Runtime hot-swapping (0.16s swap latency), preventing catastrophic forgetting (BWT +0.0008).
    • Drove benchmark classification accuracy from a 27.53% baseline to 65.70% post-sync (+38.17% gain) across a 4,752-image testbed while maintaining non-negative backward transfer (BWT +0.0008).
  • Stack: Python, PyTorch, ONNX Runtime, YOLOv8, MQTT (Mosquitto/Paho), Gaussian KDE.

3. HiJack: Smart Bio-Telemetry Wearable System

Status: Completed (National Research Grantee - PKM-KC)

  • The System: Wearable hypothermia early-warning system engineered for extreme outdoor conditions.
  • Engineering Highlights:
    • Secured competitive national funding from the Ministry of Higher Education & Research (PKM-KC 2025).
    • Authored and registered 2 National Intellectual Property Rights (HAKI).
    • Built real-time bio-sensor data acquisition and statistical noise-filtering pipelines for telemetry streaming.
  • Stack: ESP32, Embedded C++, Sensor Fusion, Telemetry Protocols.

💻 Technical Toolbox

  • Languages & Firmware: C++, C, Python, JavaScript, FreeRTOS, Arduino/PlatformIO.
  • Edge AI & Machine Learning: PyTorch, ONNX Runtime, Ultralytics YOLOv8, TensorFlow Lite, Prompt Engineering (Dicoding Certified ID: 4EXG1W3D1PRL).
  • Protocols & Systems: MQTT (Mosquitto/Paho), LoRa (SX1278), TCP/IP, REST APIs, Linux/Zsh, Git.

🏆 Key Honors & Certifications

  • National Research Grantee (PKM-KC 2025): Ministry of Higher Education & Research.
  • 1st Place & Theme Winner: IoT for Health & Safety — ISSIC 2025.
  • Best Technology Award: PBL EXPO 2025 (Politeknik Negeri Malang).
  • Prompt Engineering for Software Developers: Dicoding Indonesia (Google Authorized Partner).
  • Bronze Medalist: National Essay Competition — Forum Indonesia Muda (FIM) 2 (2026).

Pinned Loading

  1. Template-Research Template-Research Public

    Python 5