Khaing Naing

ML engineer working on computer vision and clinical AI — I build the full stack from training pipelines to production inference.


2024 -

I work as a Machine Learning Engineer at BrainEye, a brain-health company building clinical eye-tracking technology for neurological assessment. My main project was researching and deploying a 27-landmark heatmap eye-tracking model to replace an older dlib 27-landmark approach, achieving a 45% reduction in mean landmark error and halving the failure rate on clinical benchmarks. To get there, I built the full retraining pipeline from scratch: data augmentation (including synthetic occlusion and motion blur to handle real-world conditions), subject-aware train/test splitting to prevent identity leakage, PyTorch model training, and TFLite export for mobile deployment. I also integrated the model into the production C++ inference pipeline and built the evaluation tools used to validate it across clinical datasets.

More details soon.

May-Aug 2024

Back to UNSW, worked as a Research Assistant on NLP for sentiment and sarcasm classification. My work focused on data collection/preprocessing and creating a benchmark across different English dialects; the project is funded by a Google Research Scholar Grant and led by Dr Aditya Joshi.

2024

At the start of 2024, I participated in the Top100 Future Leader Award hosted by GradConnection and was selected as a finalist for the Rio Tinto IT Award.

2023-2024

I continued on the research path and joined CSIRO as a research intern supervised by Dr Necva Bolucu and Dr Maciej Rybinski and Dr Huichen Yang, where I evaluated re-scoring methods (e.g., PICK) to reduce hallucinations and improve relevance. This work deepened my NLP focus on knowledge-grounded conversational agents—exploring retrieval baselines, inference-time re-ranking, and faithfulness/groundedness evaluation.

Jan-Dec 2023

For my thesis, I explored deep learning for design: an intensive literature review and exploratory experiments on ggenerative ML for design—encoder–decoder, VAEs/GANs, diffusion models —sponsored by UNSW School of Computer Science and Engineering and Automi Design. Supervised by Professor Helen Paik and Professor Salil Kanhere.

2023-2024

I became a University Ambassador through UNSW’s outreach program — supporting students from low-socioeconomic backgrounds through workshops and mentoring, connecting them with resources to plan their futures.

2022-2023

Awarded a Taste of Research scholarship at the University of New South Wales, where I worked with Prof. Chun Tung Chou on ionic polarisation in capacitors to develop non-destructive methods for milk-quality testing.

2019-2024

BEng (Honours) at the University of New South Wales. It’s where I first got into coding.

Projects
AgeVision, this project uses a combination of Convolutional Neural Network (CNN) and a regression model to estimate people's age based on various observable features from facial images. At the time this model was deployed and accessible via a website that i wrote but now it has been taken down to improve model accuracy.
NutriVision (Just Started) is a project using Computer Vision to automate calories estimation from images and recommend basic nutrition advice.
Publications & coauthors
ALTA 2023 Shared Task
Duke Nguyen, Khaing Myat Noe Naing, Aditya Joshi