Centaurus
A personal AI assistant and robot, developed for voice with software prototypes and Nvidia hardware.
AIHardware
View project →
THE WORKBENCH OF AKASH JAYAPRABALAN
Mechanical / EE / CS
I’m Akash, a Mechatronics Engineering student at Unimelb. I bring software, electronics and physical systems together to build intelligent products.
THE INTERSECTIONAJ
HARDWARE
SOFTWARE
INTELLIGENCE
Click a layer to filter.
IDEAS INTO SYSTEMS. SYSTEMS INTO THINGS.
SCROLL TO EXPLORE ↓
01 / SELECTED WORK
Ideas, iterations
Ideas, iterations
and things that work.
Physical prototypes. Software products.
The work, the constraints and the solution.
Farmer’s Intuition
A machine learning prototype that turns weather, soil and humidity data into irrigation recommendations, with text and voice output.
AISoftware
Watch film · 0:23
Joynex
A student connection platform for exploring activities, groups and the first step toward meeting people.
Software
Watch film · 0:25
Model Council
An AI application with Solo, Compare and Council/Upshot workflows, built around model interactions and the systems that support them.
AISoftware
Watch film · 0:24
Nemotron 3.5 ASR
A speech-to-English software experiment connecting Nemotron 3.5 ASR, NLLB-200 translation and macOS speech output.
AISoftware
Watch film · 0:23
NOK
Four iterations of a handheld behavioural support concept using cooling, vibration and light. Paused when the design fell short.
Hardware
View project →
OpenCV Face Check-In
A local face recognition demo connecting enrollment, multi-face scanning and check-in records through OpenCV, FastAPI and SQLite.
AISoftware
Watch film · 0:22
Smart pill box
A three-person mechatronics build: six compartments, scheduled alerts and load-cell estimates. I own the display and control system.
HardwareSoftware
View project →
Smart voice recorder
An ESP32 audio experiment evolving from microSD recording toward buffered streaming and host side WAV handling.
HardwareSoftware
View project →
Trade
Python-based research into stock-price behaviour during the first 30–60 minutes after market opening.
Software
Watch film · 0:23
YOLO Mac Vision
A Python object detection tool for Apple Silicon Macs, with image, video and webcam workflows, CPU/MPS benchmarking and graceful fallbacks.
AISoftware
Watch film · 0:25
Centaurus
01A voice. A response. Closer to the edge.
A personal AI assistant and robot, developed for voice with software prototypes and Nvidia hardware.
WHAT IT TAUGHT ME
Build the subsystems. Understand their limits. Then bring them together.
AIHardware
Explore project →
Farmer’s Intuition
07From field data to a clearer decision.
A machine learning prototype that turns weather, soil and humidity data into irrigation recommendations, with text and voice output.
WHAT IT TAUGHT ME
A model’s output becomes more useful when someone can understand it.
AISoftware
Explore project →
Joynex
04Making room for real connection.
A student connection platform for exploring activities, groups and the first step toward meeting people.
WHAT IT TAUGHT ME
Finding your people starts with knowing where to join in.
Software
Explore project →
Model Council
03One question. More perspectives.
An AI application with Solo, Compare and Council/Upshot workflows, built around model interactions and the systems that support them.
WHAT IT TAUGHT ME
The wait before a model responds can begin long before the model is called.
AISoftware
Explore project →
Nemotron 3.5 ASR
09From speech to English. On a Mac.
A speech-to-English software experiment connecting Nemotron 3.5 ASR, NLLB-200 translation and macOS speech output.
AISoftware
Explore project →
NOK
02Rethinking the habit. Through hardware.
Four iterations of a handheld behavioural support concept using cooling, vibration and light. Paused when the design fell short.
WHAT IT TAUGHT ME
A prototype can teach you when to stop and what a redesign must solve.
Hardware
Explore project →
OpenCV Face Check-In
10From a face in the frame to a local check-in.
A local face recognition demo connecting enrollment, multi-face scanning and check-in records through OpenCV, FastAPI and SQLite.
WHAT IT TAUGHT ME
Recognition is one step. Enrollment, unknown faces and repeat scans need a clear workflow too.
AISoftware
Explore project →
Smart pill box
06A little more clarity. Every day.
A three-person mechatronics build: six compartments, scheduled alerts and load-cell estimates. I own the display and control system.
WHAT IT TAUGHT ME
A change in mass can suggest removal. It cannot tell you a dose was taken.
HardwareSoftware
Explore project →
Smart voice recorder
05From sound to a saved signal.
An ESP32 audio experiment evolving from microSD recording toward buffered streaming and host side WAV handling.
WHAT IT TAUGHT ME
Move the audio early. Keep the embedded device focused on acquisition.
HardwareSoftware
Explore project →
Trade
08A market hypothesis. A data workflow.
Python-based research into stock-price behaviour during the first 30–60 minutes after market opening.
WHAT IT TAUGHT ME
Treat a strategy as a question to test, not a result to promise.
Software
Explore project →
YOLO Mac Vision
11Local vision. Measured on the machine that runs it.
A Python object detection tool for Apple Silicon Macs, with image, video and webcam workflows, CPU/MPS benchmarking and graceful fallbacks.
WHAT IT TAUGHT ME
Measure the device choice, and make the fallback part of the design.
AISoftware
Explore project →
It started in
a hardware shop.
It kept growing.
My first workbench was my father’s hardware shop in Sri Lanka. Tinkering, soldering and repairing things gave me a practical way into engineering: take something apart, understand it, then see what you can make it do.
I moved to Melbourne to study Mechatronics Engineering at the University of Melbourne. From the beginning, I’ve been interested in connecting software with sensors, electronics, actuators and the physical world.
I use AI throughout research, coding, debugging and documentation. The work still comes back to testing: does the signal make sense, does the part fit, does the interaction actually work?
I’ve worked to help fund components and experiments. Building within a personal budget has made me resourceful and more willing to change direction when the evidence calls for it.
EDUCATION
Bachelor of Science · Mechatronics Engineering Systems
University of Melbourne · 2024–present
LANGUAGES
English
Tamil
Sinhala
03 / EXPERIENCE
Professional
July 2026–present
NOW
LAF Technologies Pty Ltd
Calibration Laboratory Technician
Calibration, verification and testing of measurement instruments. I investigate out-of-tolerance results and instrument faults, carry out repairs within my assigned responsibilities, and maintain calibration records, certificates and measurement traceability.
After a senior technician left, I took on greater responsibility, learning quickly while becoming more dependable in a practical technical environment.
Engineering teams
April 2026 — Present
NOW
Software Engineer, Humanoid Robot Team
Melbourne Space Program
Local computer vision on NVIDIA Jetson Xavier NX for humanoid robotic arm movement and water glass pickup. Vision to action integration across perception, arm control and embedded hardware/software systems.
February 2025–July 2025
Machine Learning Engineer
Aerospace and Rocket Engineering Society
Data preprocessing, feature engineering and lightweight machine learning approaches for flight telemetry on constrained hardware.
July 2024–January 2025
Systems Engineer
Melbourne University Racing
Temperature sensor selection and calibration, sensor interfacing and electronics validation for an electric Formula SAE vehicle.
04 / HOW I WORK
Across the stack.
Across the stack.
Close to the problem.
Signals & physical systems
Sensor integration, embedded audio, analog measurements and firmware logic. Grounded in projects where software has to work with a physical input.
ESP32 · C/C++ · Linux · sensors & actuators
See the pill-box firmware work →Applications & AI
Interfaces, API integration and the supporting request, account and payment flows. AI-assisted development is part of my research, coding and debugging process.
React · TypeScript · Node.js · SQL · Supabase
Explore Model Council →Data & measurement
Calibration and traceability in the lab, data preprocessing and model experiments in projects. Both start with understanding the input before trusting the output.
Python · pandas · NumPy · scikit-learn · MATLAB
Explore Farmer’s Intuition →Skills
Fabrication
- CNC milling
- Metal shop
- Wood shop
- Wood lathe
- Laser cutting
- 3D printing
Electronics & making
- Electronics & soldering
- Reflow soldering
- General training
Workshop & lab safety
- Health & safety
- Laboratory safety
- Manual handling
- Chemical management
- Plant safety
- Personal protection
- Computer workstation safety
05 / BEYOND THE BUILD
Engineering takes people.
Organising, listening and bringing people together have been part of my work too.
Interact District 3220
Zonal Head
Across four years of involvement in Interact, I worked with student leaders on community activities and club coordination. As Zonal Head, I supported the administration and coordination of more than 30 clubs across three districts, helping club leaders communicate and keep shared initiatives moving. The experience shaped how I work with people: listening to different perspectives, coordinating across teams and following through on responsibilities.
Senior Peer Mentor
Melbourne Peer Mentor Program - Unimelb
Over two years in the University of Melbourne’s Peer Mentor Program, I supported first-year students as they settled into university life, stepping into a Senior Peer Mentor role in my second year. It reflects something I enjoy beyond building: helping people find their footing, make connections and grow in confidence.
Mtech Consulting
Events Director
Professional events, workshops and networking, with responsibility for logistics and stakeholder communication.
Circula
B2B Associate Intern
I supported university-club outreach and launch preparation, contributed to campaigns across three social platforms, and collaborated with growth, product and operations teams.
OFF THE CLOCK
Away from the bench, I enjoy chess, poker, long-distance athletics and making videos. Chess has been a lesson in persistence: improving through competition eventually led me to become captain.
WHAT’S NEXTMy next step is practical engineering work across robotics, electronics and software. Longer term, I’m interested in robotics research, building useful products and sharing what I learn.
06 / START A CONVERSATION
MELBOURNE · --:--
Good things start
with “what if?”
Robotics, an early idea, or something
that doesn’t fit neatly in a box.
Let’s build something worth making.