Kaggle Grand Master | Data Scientist | AI Engineer
Building agentic AI systems, fine-tuning LLMs & VLMs, and designing financial-crime AI at the Commonwealth Bank of Australia.
Data scientist and AI engineer turning frontier research into production-ready systems. At the Commonwealth Bank of Australia I architect agentic AI frameworks for financial-crime investigation — orchestrating sub-agents, MCP tools, and memory-engineered pipelines at scale.
My work spans LLMs, VLMs, RAG, differential privacy, and evaluation science. Researched at Harvard, MIT, and Carnegie Mellon. Kaggle Grandmaster with 400,000+ views.
Architected an end-to-end agentic AI framework with specialised sub-agents and MCP tools as RESTful APIs. Engineered multi-agent pipelines using episodic & semantic memory stores. Designed LLM-as-a-Judge evaluation suite.
Contributed to first enterprise-scale AI models for financial-crime detection. Applied differential privacy for privacy-preserving AI over sensitive data.
Fine-tuned VLMs & LLMs for document fraud detection. Built agentic AI for change & migration risk assessment.
NLP with LLMs for medical information structuring. Built retrieval pipelines using PubMed and medical-domain sources.
Prompt engineering and LLM fine-tuning. Contributed to RLHF methodology for model responsiveness.
Led ML modelling for Acute Lymphoblastic Leukemia detection — 95% classification accuracy.
Biases in LLM text generation across linguistically diverse medical contexts.
ML & deep learning for subtomogram identification in Cryo-electron Tomography.
AI/ML for statistical modelling, thermodynamics, and bio-inspired algorithms.
ML algorithms for IoT security breach detection.
Diffusion models across diverse domains.
Fine-tuning pre-trained computer-vision models — ImageNet, ResNet.
St. Martin's Engineering College
European Open University
Open to research collaborations, agentic-AI projects, and conversations about applied ML.