Skip to main content
Swin Ratnayake

Swin Ratnayake

Data Scientist | AI Engineer

Talent
Hamilton, OntarioMember since March 3, 2025
·

Hire this person

Send a job offer directly to this candidate

About

Experienced in AI-driven solutions, LLMs, NLP, and computer vision across healthcare, robotics,

recruitment and digital marketing. Skilled in Python, R, SQL, and cloud platforms, with a strong focus on optimizing AI infrastructure and automating processes. Passionate about solving complex problems with data and machine learning.

Experience

HYPE Invention Pty Ltd (2023 - Present)

  • AI-Powered Photographer Robot: Developed an autonomous photography AI using Florence-2, OpenAI CLIP, and OpenCV for scene analysis, object detection, and aesthetic evaluation. Integrated AI with robotics for precise camera control.
  • AI-Assisted Patient Management System: Fine-tuned Mistral NeMo with medical data, outperforming GPT-4o in domain accuracy. Deployed on Google Cloud Vertex AI and Kubernetes, reducing infrastructure costs by 88% ($35K → $4K/month).

Medfuture Healthcare Pty Ltd (2021 - 2023)

  • Built SQL pipelines and Power BI dashboards for healthcare analytics.
  • Developed machine learning models to automate recruitment, improving candidate matching by 80%.
  • Conducted digital marketing analysis to optimize campaign performance.

Lanka Hospitals PLC (2018 - 2021)

  • Performed microbiological & molecular-biological lab analyses.

Key Skills:

Computer Vision, NLP, Machine Learning, AI Model Optimization, SQL, Python, Power BI, Cloud AI (Google Cloud, Kubernetes), Robotics AI.

Education

Education:

  • Master of Science in Data Science – Cardiff Metropolitan University, UK (2023)
  • Bachelor of Science in Biological Science – State University of New York at Oswego, USA (2018)

Key Project:

Lung Cancer Prediction Web App (Master’s Thesis – ML & Healthcare AI)
  • Developed a machine learning-based web app for lung cancer risk prediction using Flask and scikit-learn, deployed on AWS.
  • Trained linear regression and random forest models on a dataset of 535 individuals, achieving 97% AUC and 0.76 R², outperforming benchmarks.
  • Enhanced early detection and clinical decision-making for high-risk patients.

Reviews

Related