AI Engineer

Models that
behave in the
real world.

LLMs · RAG · NLP · Applied ML

I'm an AI engineer who likes getting models to actually work outside the notebook. With a Master of Artificial Intelligence behind me, I've built chatbots that stay on the facts, translation that runs in real time, and APIs that hold up when real people start using them.

See the work Get in touch ›

The hard part isn't building the model. It's getting it to behave when real people show up.

0
Lyrics translated across 100+ languages
0
Peak BLEU translation quality
0
Inference latency per line
0
AI/ML projects shipped
Selected work

A closer look.

Retrieval-Augmented Generation

A chatbot that only speaks from your sources.

It crawls a website, cleans the pages into Markdown, and embeds them so every answer comes straight from the source. I added chunking, reranking and guardrails so it stays on topic — and is honest when it doesn't actually know.

PythonOpenAI APIspgvectorPlaywrightREST APIs
What's your refund window?
Refunds are available within 30 days of purchase, provided the item is unused.
↳ grounded in /policies/returns
And for opened items?
Opened items qualify for store credit within 14 days.
↳ grounded in /policies/returns#opened
NLP · Translation Pipeline

10,000 lyrics. 100+ languages. Sub-second.

I built this one end to end on Meta's M2M100 — pulling lyrics in, cleaning and chunking the text, running translation, checking the output, and storing it. It answers in real time through a Flask API behind a Vue.js frontend.

87%Peak BLEU
<0.4sPer line
100+Languages
M2M100Hugging FacePyTorchFlaskVue.jsMySQL
EN — source
ES / FR / JA / HI …
Reinforcement Learning

Symbolic planning meets Q-learning.

A Pac-Man Capture-the-Flag agent built in UC Berkeley's AI framework. I mixed hand-written symbolic logic with feature-based Q-learning so it could plan routes and make smarter calls against a live opponent.

PythonQ-LearningSearchReinforcement Learning
Generative AI · Computer Vision

Synthetic histopathology, expert-grade.

Research into using GANs to generate synthetic histopathology images. The workflows I trained produced samples that domain experts rated 80% indistinguishable from the real thing — and I worked closely with medical staff the whole way to get there.

PyTorchTensorFlowGANsComputer Vision
Geospatial Analytics

Road-crash data, made explorable.

An R Shiny app for digging into Victoria's road-crash data. I cleaned the geospatial data with dplyr, charted the patterns with ggplot2, and mapped severity with Leaflet so you can explore where and how crashes happen, by hand.

RR Shinydplyrggplot2Leaflet
Data Visualization

A season of F1, at a glance.

An interactive F1 dashboard in R for comparing race results, driver form, and how a whole season played out — built for people who like their racing with a side of stats.

RDashboardingStatistical Analysis
Career

Experience

Feb 2025 – Jul 2025

Machine Learning Engineer

Monash University — Faculty of Information Technology
  • Built and deployed the multilingual lyrics translation system on Meta's M2M100 (10,000+ lyrics, 100+ languages).
  • Brought inference under 0.4s per line via regex cleaning, batching and model tuning; reached peak BLEU of 87%.
  • Exposed the model through Flask endpoints wired into a Vue.js frontend serving real-time requests.
Aug 2024 – Feb 2026

Data Scientist & Digital Media Specialist

ITCC — Digital Media & Data Analytics
  • Built Python/SQL pipelines consolidating Shopify, website and paid-media data into clean, queryable sources.
  • Built segmentation models with clustering and Scikit-learn, contributing to a 33% lift in targeting precision.
  • Cut recurring reporting errors by 35% and shipped recurring stakeholder dashboards.
Dec 2022 – Jun 2023

Research Intern — Generative AI & Computer Vision

Navrachana University — Department of Information Technology
  • Researched GAN-based synthesis of medical histopathology images.
  • Produced imagery rated 80% indistinguishable from real samples by domain experts.
May 2022 – Jul 2022

Lead Project Intern

Arth Technology — Software Automation
  • Wrote modular Python scripts automating keyword analytics and search workflows.
  • Parsed, cleaned and structured web-sourced data for downstream reporting.

Like what you see? Say hello.

Master of Artificial Intelligence, Monash University · Clayton, VIC · Full working rights in Australia.