Linh Tuan Ung

Linh
Tuan Ung

Machine Learning Engineer, Vocare Speech

  1. Vocare Speech Speech models
  2. Aformity Data onboarding
  3. Honours research Space debris
  4. GoMicro Seed forecasting

About

Adelaide, South Australia

I'm a machine learning engineer in Adelaide. My main work is speech, at Vocare Speech. Alongside it I build software at Aformity, and before both I did First Class Honours in computer vision at the Australian Institute for Machine Learning.

Also
Software Engineer Aformity
Studied
BCompSc (Hons), First Class Adelaide University
Based
Adelaide South Australia

Work

Industry and research

Vocare Speech

Machine Learning Engineer

Apr 2026 – now Adelaide

Working on the speech models behind Vocare’s pronunciation feedback, for skilled migrants in the care sector.

  • Built a full-stack internal platform (Python, TypeScript) for audio annotation and machine learning evaluation.
  • Developed scoring algorithms and data pipelines to analyse audio and classify errors.
  • Maintain large-scale datasets, with automated testing on the core analysis pipeline.
  • Automated system deployments using Docker and systemd.

Python TypeScript Docker systemd

vocare.com.au
Fig. 1 Made-up takes of a phrase as terrain: time runs back, pitch across, loudness up. Rough stretches get marked in orange, and the next take comes out smoother there. Not Vocare’s method, model or data.

Aformity

Software Engineer

Sep 2025 – now Adelaide · hybrid

Customer data onboarding: turning messy customer files into validated, import-ready data before a SaaS launch.

  • Implementation and customer success teams bring in whatever the customer has: spreadsheets, competitor exports, legacy system dumps. Aformity maps it, cleans it, validates it and gets it signed off before import day.
  • I deliver full-stack features across the web app and backend.
  • I work on reliability and developer productivity through testing and automation.
aformity.com
customers_export.csv · 6 rows Uploaded
A made-up customer file being prepared for import
# Company account.name email account.billing_email Plan plan.tier start date contract.starts_at status
1 saltbush dental pty ltd Ops@Saltbush.com.au growth 03/07/26
2 KITE LOGISTICS billing@kite.io Scale 2026-07-09
3 Ferro Labs team 9 Jul 2026
4 kite logistics BILLING@KITE.IO scale 2026-07-09
5 Palm & Pine Studio hello@palmpine.co Enterprise 15/07/2026
6 tidewater health admin@tidewater.health GROWTH 2026-07-21
  • trim + title case
  • trim + lowercase
  • normalise plan
  • parse date → ISO
  • dedupe by email
Fig. 2 Step through it, or fix row 3 yourself by typing an email. Made-up data: not Aformity's interface or code.

Australian Institute for Machine Learning

Honours Research Student

Jul 2025 – Jun 2026 Adelaide

Counting the fragments of a satellite breakup in simulated imagery, where most pieces are too small and too crowded for a detector.

  • Generated a multi-camera synthetic dataset of a debris cloud, built in Blender and scripted in Python.
  • Measured where a standard object detector (YOLOv8) stops coping with small, densely packed fragments.
  • Applied density estimation (CSRNet) as a different way to count, and tested whether it held up on camera viewpoints it had never seen.

Python PyTorch Blender YOLOv8 CSRNet

Fig. 3 A satellite breakup, simulated in your browser and modelled on the renders from the research. The lens counts what sits under it two ways: the boxes a detector could draw, and a density map. Toy data, not thesis results.

GoMicro

AI Research Intern

Dec 2025 – Jan 2026 Tonsley, SA

Seed growth forecasting with computer vision, deployed on edge hardware for an international agricultural export client.

  • Lead engineer for deploying edge AI computer vision systems on NVIDIA Jetson hardware.
  • Developed and optimised deep learning pipelines that automate seed growth forecasting, cutting down manual assessment cycles.
  • Managed the full lifecycle from research to deployment, including client requirements and hardware integration.

Computer vision Object detection NVIDIA Jetson

tray_03 · 40 seeds hour 00 · 00 / 40 sprouted
Fig. 4 Drag through a made-up time-lapse: seeds germinate, get boxed as they do, and the forecast firms up as the hours pass. Not GoMicro's data, model or client.

Experience

Since 2024
  1. Jul 2026 – now

    Vocare Speech Machine Learning Engineer

  2. Apr – Jun 2026

    Vocare Speech Machine Learning Engineer Intern

  3. Sep 2025 – now

    Aformity Software Engineer

  4. Dec 2025 – Jan 2026

    GoMicro AI Research Intern

  5. Jul 2025 – Jun 2026

    Australian Institute for Machine Learning Honours Research Student

  6. Feb – Jul 2025

    Australian Institute for Machine Learning 3D Modelling Intern

  7. Dec 2024 – Feb 2025

    Lyra Technologies Full Stack Developer Intern

  8. Nov 2024 – Jan 2025

    University of Adelaide Summer Research Student

  9. Aug 2024 – Mar 2025

    Budi Cloud Backend Engineer

Education

  • Bachelor of Computer Science (Honours), First Class

    Adelaide University · 2025 – 2026

    Thesis: fragment counting in simulated satellite breakup imagery using density estimation.

  • Bachelor of Computer Science, Artificial Intelligence major

    University of Adelaide · 2022 – 2025

Recognition

  • Global Citizens Scholarship

    University of Adelaide · 2022, 2025

  • ICPC South Pacific Preliminary

    Competitor · 2024

  • National Award for Excellence in Informatics

    Vietnam · 2022

Community

  • Open Source Developer

    Adelaide University Computer Science Club · 2024 – now

  • Content Contributor

    LQDOJ online judge · 2022 – 2023

Toolkit

Machine learning
PyTorch, OpenCV, Object detection, Density estimation, Blender (bpy)
Software
Python, TypeScript, Go, C++, SQL
Systems
Linux, Docker, Git, PostgreSQL, AWS

Contact

Email is best

Happy to talk about machine learning, computer vision, or anything on this page.