I measure systems first, then build the models and tools the measurements ask for.
Who is looking
Claim
I’m Adhi, Head of AI Architecture & Research Analysis at Ozler Care Solutions in Melbourne. I lead the architecture of a compliance platform for Aged care and NDIS providers. From January to June 2026 I was also a Technical Business Analyst at Helloworld Travel Limited (ASX:HLO). Earlier: StaffGrid and data work at Zenuca, research analysis as an intern at Deakin’s A²I², RPA work at Visualyze.AI in Trivandrum. I build independent AI systems, analyse businesses and data, contribute to research on LLM evaluation and model routing at A²I², and publish essays on AI. I choose the simplest architecture that works. I derive intelligence from data. In all of it: making complexity useful.
Also people call me Adhi. Pure coincidence that it spells out my job:
- A Architecture.
- D Data.
- H Hypotheses.
- I Intelligence.
The dated record
Feb 2026topresent
Ozler Care Solutions (Ozler Tech Pty Ltd), Melbourne
Head of AI Architecture & Research Analysis
A compliance, training and workforce-credentialing platform for aged care and NDIS providers, on Azure: its Solution Architecture Specification, the build of its monorepo, and AI-assisted serious-incident reporting and policy generation with human review before anything reaches a regulator.
The platform’s data, security and identity architecture, and the research analysis of aged care and NDIS regulation and the provider market behind its product line and launch.
Jan 2026toJun 2026
Helloworld Travel Limited (ASX: HLO), South Melbourne
Technical Business Analyst & Product Engineer, Wholesale Business Systems
Technical business analysis, including the Tourplan WebConnect implementation for the inbound brands, the 8x8 webchat scope and a customer-facing chatbot on 8x8; LogLens, an Nginx and HAProxy log parser the team adopted for daily use; a Power BI CEO dashboard for MyWay Travel & Events, GA4 analytics pipelines with automated executive reports, and a strategic financial analysis of ASX:HLO for board presentation.
Dynatrace observability: monitoring, dashboards, alerts and SLOs, with their configuration automated through the Dynatrace API; a configuration audit that cut the annual cost; root-cause analysis of recurring booking-integration errors; an observability assessment of the full estate for the CTO.
Jan 2024toMay 2025
Zenuca Technologies, Melbourne
Business Data Analyst & AI/Automation Developer
Data, automation and software delivery for a healthcare and personal care staffing provider. StaffGrid, a healthcare staffing system in Blazor WebAssembly and MudBlazor with role-based dashboards and tier-based shift broadcasting to agencies: I led its design and development, produced its architecture documentation and ran design sessions with an overseas development team.
Power BI dashboards for staff allocation, coverage and compliance; demand forecasting for healthcare workers; NLP applied to client feedback; Power Platform automations, custom API integrations, and compliance, certification and licence records.
Sep 2024toDec 2024
Applied Artificial Intelligence Institute (A²I²), Deakin University, Burwood
Research Intern, AI & Data Science
YouSay-WeSay, a research analysis project on global perceptions of the 2024 US election, where I was the analyst: I processed the multi-country survey data and built Python (matplotlib, seaborn) and JavaScript visualisations. The research was featured in the London School of Economics USAPP blog.
Jul 2023toDec 2023
Regis Aged Care, Melbourne
Office Coordinator
Feb 2022toJan 2024
Visualyze.AI, Trivandrum, India (hybrid)
Associate Software Engineer (Part-time)
RPA automations for data processing; Python and C# pipelines; code reviews.
Jul 2022toJul 2025
Deakin University, Burwood
Bachelor of Computer Science (Data Science), Distinction (WAM 77)
Generative AI Advisory Board Member; AI Research Team Member, A²I²; Student Ambassador.
Co-founder of The Malayali Association of Deakin Inc. (Deakin Mallus), an incorporated cultural association
Certificates: Python for Data Science and AI (IBM); Introduction to AI (IBM); Generative AI and Prompt Engineering; Technical Support Fundamentals (Google); a Power BI job simulation (PwC); a data visualisation job simulation (TCS).
Three proof lines
−50 % data processing time, with RPA automations, at Visualyze.AI. Reported
SkillMate, an AI platform for migration and study-abroad businesses: I designed and built it. Built
−16 % annual Dynatrace cost, in my first week at Helloworld, through a configuration audit. Measured
roots. English, Malayalam and Tamil.
Selected work
Five pieces of work done in roles, nine of my own builds and three research projects. Each opens on its own page with what it is and where it stands.
- Ozler Care Solutions
- Helloworld Travel
- Zenuca Technologies
- A²I², Deakin
- Regis Aged Care
- Visualyze.AI
- Deakin University
- LSE USAPP
Where I have worked and studied, in the order of the dated record, and the one write-up that acknowledges me.
Essays and notes
| Title | About | Date |
|---|---|---|
| 2026 | ||
| The First RecallAlso on Medium | When the U.S. Government forced an AI model offline worldwide. | |
| Apple’s $1 Billion AdmissionAlso on Medium | When the world’s most valuable company gives up on building AI. | |
| The Trillion-Dollar Safety PremiumAlso on Medium | How the Anthropic–Pentagon standoff created $585 billion in market value. | |
| The Agent as Economic ActorAlso on Medium | When AI spends money, who owns the consequences? | |
| The Liability Vacuum in Agentic AIAlso on Medium | Why tort law breaks when the agent has no principal. | |
| The credible commitment problem in AI safety: lessons from the Anthropic–Pentagon standoff | A game-theory reading of a refusal that accepted blacklisting rather than capitulate. | |
| The $53 Billion Blind Spot: Why AI Agents Will Eat Travel Distribution From the Inside Out | And why the companies that control the pipes, not the interfaces, will win the next decade. | |
| They’re Not Coming For Your Job. They’re Coming For Your Purpose. | OpenClaw launched a mirror, and what stared back was uncomfortable. | |
| The Encoding Gap: How AI Is Reshaping the Architecture of Human Expertise | What happens to human knowledge when the machines that augment it begin to outperform the experts who built it? | |
| 2024 | ||
| Research proposal: advancing Random Forests for high-dimensional, imbalanced data in biomedical applicationsA note, on Medium | A proposal, not a paper: an enhanced Random Forest approach for imbalanced biomedical data. | |
Claim
Source: Substack, cross-posted to Medium. Date: Feb to Jun 2026; one note from Oct 2024. Scope: Titles and dates as published. Links go to the originals. Essays are essays and notes are notes; nothing here is called research.
The essays are about what AI is doing to work, to markets and to responsibility. I publish them on Substack, with a copy on Medium, and the dates here are the real ones. Published
Write about analyst, data-science or AI-engineering work, in Melbourne or remote.
You looked closely. So do I.