Thinking ahead of the technology.
Cofounder and CTO at IdeaBoxAI, leading a 50 person engineering organisation.
Published researcher in AI. I have built the engines, from vector databases to neural network architectures, and used them to change how real industries work. I speak and write about where AI and technology are really going, with the data to back it up.
Bootstrapped
2M+
revenue at IdeaBoxAI, grown to 35+ people
Published
2 papers
in IEEE Xplore and Applied Computer Systems
Speaker
Bangkok
Responsible Seafood Summit 2026
01 The argument
Most companies have an AI story. Few have an AI strategy.
The board wants a direction, the team wants a roadmap, and the demo works right up until it meets real data, real security reviews and real people. I work on the thinking first, then the research and the engineering that make it hold.
The problem
AI decisions made from headlines rather than evidence, pilots that never earn a security sign off, and engineering teams asked to adopt tools nobody has shown them how to use well.
What I do about it
Lead the engineering organisation that ships Persona Intelligence, our platform, into regulated industries, publish research on making models trustworthy, and put the data in front of leaders and teams so the decisions get made on evidence.
02 Proof
From the engine room to the industry
I have built at both ends: the infrastructure most people never see, and the outcomes a business actually notices. The range is deliberate. Deep technology only earns its keep when it changes something real.
Engines
Built the engines
- 01A vector database engine, optimised for memory and scale
- 02Neural network architectures, including explainable models for medical imaging
- 03Memory systems that let AI agents work at billion vector scale
- 04An enterprise AI platform with connectors, security and observability built in
Outcomes
Changed the outcomes
- 01Sustainability reporting that enterprises can produce and stand behind
- 02Governance, risk and compliance work done in a fraction of the time
- 03A seafood industry and its supply chain moving onto digital rails
- 04Sales and solutions teams that close faster with an AI copilot
Before IdeaBoxAI: six years of engineering, from apprentice to technical lead, at Mitra Innovation and RightData.
03 Where I sit
Where the three meet
- Research tells you what is possible.
- Product tells you what is worth building.
- Running a company tells you what survives.
I work where the three meet.
Most people in AI do one of these. The value I bring is doing all three at once, so the research informs the product and the product funds the research.
Entrepreneurship
Cofounder and CTO of IdeaBoxAI, an enterprise AI company headquartered in Cary, North Carolina. I lead a 50 person organisation across platform, forward deployed engineering and delivery. The platform is ISO 27001 certified and SOC 2 Type II compliant, and takes AI from first pilot to production in regulated industries.
AI research
Published work on explainable AI for cardiac MRI segmentation, in Applied Computer Systems and in IEEE Xplore. My MSc thesis at the University of Moratuwa designs a multi index memory architecture for AI agents built for billion vector scale.
Product management
I own product strategy and technical direction: what we build, for which roles, and in what order. Persona Intelligence, Helm for sales teams and Marea for the seafood supply chain all started as a product thesis before they became code.
04 Research
Research
Explainable AI, medical imaging, and how AI agents should remember.
Applied Computer Systems, vol. 30, no. 1, pp. 12–20
January 2025
Peering into the Heart: A Comprehensive Exploration of Semantic Segmentation and Explainable AI on the MnMs-2 Cardiac MRI Dataset
Mohamed Ayoob, Oshan Nettasinghe, Vithushan Sylvester, Helmini Bowala, Hamdaan Mohideen
Segmentation at Dice scores near 90 percent, with Grad CAM and feature ablation showing which regions drive each prediction.
Read the paperIEEE Xplore
2025
An Examination of Harmonizing Multi-Vendor Cardiac MRI Data for Vendor-Neutral Image Segmentation
Peer reviewed and published in IEEE Xplore. Full citation on the paper page.
Read the paperMSc thesis, Department of Computational Mathematics, University of Moratuwa
In progress
A scalable multi index memory architecture for AI agents
Hot and cold memory paths with HNSW indexing, so agent recall stays fast and accurate at billion vector scale.
04.1 Education and awards
Building products and doing research since undergrad
2019
IIT Infoschol Scholarship
Full scholarship from the Informatics Institute of Technology, Sri Lanka, to study for a BEng (Hons) in Software Engineering awarded by the University of Westminster, UK.
2020
Apprentice engineer at Mitra Innovation
Started as an apprentice engineer while studying. The first of six years of engineering, from apprentice to technical lead.
2021
National Ingenuity Awards, SLASSCOM
Best Innovative Product / Project, for my second year product: real time streaming of patient vital signs for rural healthcare.
2024
Left Mitra Innovation as technical lead
March 2024. Moved on to consult for startups, including RightData.
2024
PROJEXPO Bronze Award, Technical Research and Development
For research on improving the efficiency of cardiac MRI semantic segmentation, the work that became the Applied Computer Systems paper.
2024
BEng (Hons) Software Engineering, First Class Honours
University of Westminster, UK, studied at the Informatics Institute of Technology, Sri Lanka, 2019 to 2024.
2024
Started IdeaBoxAI
November 2024. Founded my own company to build an enterprise agentic AI platform.
Now
IdeaBoxAI, bootstrapped to 2M+ revenue
Grown to 35+ people without outside funding.
Now
MSc, Department of Computational Mathematics, University of Moratuwa
Active research on agentic AI systems. Thesis in progress on a scalable multi index memory architecture for AI agents.
05 Writing
Writing
Long form analysis, published on this site first, with the data underneath every chart.
October 2026 · 15 minute read
Are tech workers cooked?What seven years of data say about AI and tech jobs
Seven years of labour data from the US, Europe, India and China. Interest rates started the tech jobs decline, AI is why the recovery never came, and the cost is landing on the youngest workers in every market we can measure.
Eleven charts. Every number traceable to its source.
- 12.1%fall in US information sector jobs since the November 2022 peak
- 71%of the rebound in US software developer postings went to senior roles
- 3 to 4×youth unemployment versus the headline rate in the US, India and China
Upcoming
- How an AI agent should remember: hot and cold memory pathsUpcoming
- The real cost of LLM tokens at enterprise scale, and what caching, compression and routing actually saveUpcoming
- Forward deployed engineering: how we take enterprise AI from pilot to productionUpcoming
06 Work with me
What brings you here?
Three reasons people get in touch. Each has a small first step.
Speaking
What do you speak about?
What AI is really doing to work and to engineering teams, how enterprises get from pilot to production, and explainable AI. In September 2026 I spoke at the Responsible Seafood Summit in Bangkok, and I ran HackIndia 2026 for about 300 students in Chennai. I travel from Colombo and present remotely when the time zones work.
Email me the event and the audience.
Advisory
Do you advise founders or companies?
I take a small number of advisory conversations with founders building in AI and data, usually on architecture, memory and the path from pilot to production.
Email first with what you are building.
Joining the team
Are you hiring?
We hire senior engineers who want to work on agent memory, connectors and production AI systems, and forward deployed engineers who take enterprise deployments from pilot to live.
Send a short note with something you have built.
Also asked
What does IdeaBoxAI do?
IdeaBoxAI builds Persona Intelligence, an enterprise platform for role specific AI agents with memory, skills, connectors to over 1,000 systems, and natural language workflow automation. Helm is an application on top of it for sales and solutions teams. The company is headquartered in Cary, North Carolina.
What is your research about?
Two threads. Published work on explainable AI for medical imaging, including cardiac MRI segmentation research that won a PROJEXPO Bronze Award in 2024 and became a paper in Applied Computer Systems (2025), and a paper in IEEE Xplore. And my MSc thesis at the University of Moratuwa on a multi index memory architecture for AI agents built for billion vector scale.
Can I quote or republish your writing?
Quote freely with a link back. For republishing a full piece, email me; I usually say yes with attribution and a canonical link to this site.
07 Get in touch
Get in touch.
Speaking, advisory, or joining the team. Two or three lines is enough, and I reply to almost everything.
Privacy: this site sets no cookies and collects nothing. Email goes to me and nobody else.