I build the whole stack —firmware to production software.

Real-time RTOS firmware on one end; React, Node, AWS, and neural nets built from scratch on the other.

See the work→Jad Dina · EE @ UCalgary · Calgary, AB
scroll

Right now, I'm working both ends of the stack.

Embedded2026—Present

Avionics Software Developer

SOAR — UofC Rocketry ↗
  • Designed and implemented a full data acquisition system — wrote peripheral drivers for IMU (LSM6DSO), magnetometer, barometer, and GPS, interfacing via SPI with a mutex to prevent signal coupling across channels
  • Built the flash logging pipeline: circular priority RAM buffer → sector-aligned QSPI flash writes at 200Hz, with a `bufferPerSector` strategy that erases each sector once and fills it with multiple RAM-sized pages for write efficiency
CC++STM32FreeRTOSSPIQSPIEmbeddedRTOS
Software2026—Present

Software Engineering Intern

Emly AI ↗
  • Owned the AI Analysis Pipeline project end-to-end — authored the requirements document and technical design document, then drove implementation through to delivery
  • Built the subscription-gated website chat widget system across settings UI, public embed flow, and plan eligibility logic in a TypeScript monorepo
TypeScriptReactNode.jsAWSAISaaS

The one I'm proudest of —

Python · Machine Learning · NumPy

Machine Learning Suite

91%digit recognition accuracy

A neural network with full backpropagation, written in NumPy only — no TensorFlow, no PyTorch. Part of a from-scratch suite spanning SVM, K-means, Naive Bayes, and decision trees, every model built up from the math.

Pythonscikit-learnNeural NetworksJupyter
Read the case study →
Neural network digit-recognition predictions from the from-scratch ML suite

More work

All projects→

Let's work together.

Open to internships, new-grad roles, and interesting problems.

jaddina19@outlook.com→LinkedIn ↗