AI PRODUCT DEVELOPER SYDNEY, AU

Finding signal.
Building what’s next.

I’m Jonathan Kenneth Gunawan.
I build AI products across workflows, markets, and mobile.
Currently an intern at Gradstack and an AI student at UTS.

Explore my work
SCROLL TO EXPLOREARTIFICIAL INTELLIGENCE. HUMAN CURIOSITY.
AI products & agentsMachine learningVoice & mobileFull-stack development

CURRENTLY BUILDING WITH

Gradstack®

Intern · –present

Sydney · AI & software development

Turning learning and assessment into usable software.

I contribute to Gradstack’s AI fluency assessment and learning platform as part of the engineering team.

  • Built candidate-facing assessment and results experiences, including downloadable result summaries.
  • Worked on searchable learning catalogues and interfaces that help learners find a starting point.
  • Contributed to the data models behind learning progress and achievements.

Team contributions · Public overview

SELECTED WORK

Ideas, made real.

From a question to a working system.
A few things I’ve been building.

Local platform · Private source
Flowstudio’s visual workflow editor showing a research workflow and step settings
Build, run, understand.

PROJECT 01 / AI workflows · Developer tools

Flowstudio

Product design & full-stack development

From an idea to a workflow you can inspect.

A local AI workflow platform for composing agents, tools, and reusable context in a visual editor. My work brings authoring and execution together: connect steps, run a workflow, inspect its outputs, and revisit what happened. The product includes a dependency map, durable runs, and review steps for proposed changes. It’s a working local platform, with a product walkthrough here and its source kept private.

  • Next.js
  • TypeScript
  • Python
  • PostgreSQL

Visual authoring · Reusable agents · Run inspection

Explore the project
Public research prototype · Private source
Market Cerdas public research workspace with AI chat and market research navigation
Put information in context.

PROJECT 02 / Full-stack AI · Financial intelligence

Market Cerdas

Full-stack AI development

A clearer view of the market.

An AI-assisted stock research prototype bringing market news, technical charts, and company fundamentals into one workspace. I built across the interface, APIs, retrieval, and data pipelines so a research question can lead to the context behind an answer. The public application supports ASX research; Indonesian stock prediction remains a development direction. Explore the product and its public documentation while the application source stays private.

  • React
  • Express
  • Gemini
  • Python

Market research · Grounded answers · News sentiment

Explore the project
In development · Team project
Haven AI home screen in a clean web preview with no personal journal entries
Keep your own voice.

PROJECT 03 / Mobile AI · Voice & journaling

Haven AI

Mobile & API development · Team contribution

A spoken thought. A journal you control.

A voice-first journaling product built around deliberate review and device-local storage. My contribution to the team project spans the Expo mobile app, transcription integration, editable journal drafts, and supporting API and build workflows. The experience moves from recording to a reviewed transcript and a saved entry, with AI assistance as an explicit choice. This is an in-development product showcase, with source and personal journal content kept private.

  • React Native
  • Expo
  • TypeScript
  • SQLite

Voice capture · Review before saving · Local journal vault

Explore the project

More ways I build

CONTRIBUTIONS & EXPERIMENTS

Team contribution

Serendip-TG

AI-generated story quizzes and profile-driven themes for a Telegram Mini App. My contributions include background generation, admin draft editing, and generation-status handling.

  • FastAPI
  • Celery
  • React
Read my contribution

Interactive prototype

Server-driven UI

A profile-driven interface that changes its palette, typography, and avatar from structured data. Explore the Astrana demo to see the visual side of AI personalisation.

  • React
  • TypeScript
  • LangGraph
Explore the demo

ML research · University capstone

Cryptocurrency forecasting

Forecasting research across BTC, ETH, and SOL using price, on-chain, and macroeconomic signals. Comparing statistical, machine learning, and transformer baselines with explicit evaluation metrics.

  • Python
  • Time series
  • Model evaluation
View research repository
Earlier ML work Research foundations & coursework
Financial NLP

Indonesian news sentiment

Financial news collection, event clustering, and sentiment classification. Supporting research into Indonesian markets.

View repositoryTabular ML

Weather prediction

Leakage-safe rainfall classification. Random Forest baseline: 0.86 test accuracy.

View repositoryComputer vision

Brain tumor classification

MRI classification using LBP + HOG features and SVM. Baseline test accuracy: 81.36%.

View repositoryObject detection

Skin disease detection

A reproducible YOLO training and inference workflow. Baseline mAP@50: 0.696.

View repository
Jonathan Kenneth Gunawan
THE PERSON BEHIND THE PROJECTS↗

A LITTLE ABOUT ME

Curiosity first.
Then, code.

I’m Jonathan, a Bachelor of Artificial Intelligence student at the University of Technology Sydney and an intern at Gradstack since July 2026. I’m interested in what happens when rigorous machine learning meets something people can actually use.

My work spans agent workflows, financial research, voice-first mobile experiences, and learning platforms. I enjoy connecting the model to the interface and the system behind it. I care about clear thinking, reproducible work, and responsible AI.

More about my journey
BASED IN SYDNEYSTUDYING AT UTS

MY TOOLKIT

Across the whole stack.

The tools change. The approach stays:
understand, experiment, evaluate, build.

Build the product

Interfaces and APIs that make the model useful.

  • TypeScript
  • React
  • Next.js
  • React Native / Expo
  • Node.js
  • FastAPI
  • Express
  • Python
  • SQL

Find the pattern

Models built around the problem and measured against a baseline.

  • XGBoost
  • AutoGluon
  • Chronos
  • LSTM
  • Time series

Connect the context

Language systems grounded in the information that matters.

  • LLMs
  • LangGraph
  • RAG
  • Embeddings
  • Hybrid search
  • Sentiment analysis

Data & infrastructure

Data storage and background jobs that keep products running.

  • Supabase
  • PostgreSQL
  • SQLite
  • SQLAlchemy
  • Redis
  • Celery

Always exploring — evaluation, fine-tuning, and MLOps. Building with responsible AI and reproducibility in mind.

THE PLAYGROUND

Less theory.
More “try this.”

A couple of small experiments.
An invitation to play with the possibilities.

Hand landmark tracking demonstration INTERACTIVE DEMO

COMPUTER VISION

A show of hands.

Real-time hand tracking. Your camera, a little AI, and all ten fingers.

Try hand trackingCamera required
Rock paper scissors game against an adaptive AI INTERACTIVE DEMO

ADAPTIVE MACHINE LEARNING

Your next move?

Rock, paper, scissors—with an opponent that learns your patterns.

Challenge the AINo setup needed

THE NEXT GOOD IDEA STARTS WITH A CONVERSATION

Let’s build
something thoughtful.

Say hello jonathan.kenneth.gunawan@gmail.com