2026 to present
Now
CXCaller
Co founder
An AI agent calling platform. Agents place and take the calls; the platform carries the campaign, the context and every outcome.
Chamidu Himantha/Software engineer, AI researcher
I build software that talks to people, thinks for itself, knows what to say, acts on its own, learns as it goes
I build AI agents that do real work: voice, chat, and the tools they reach for. Most of my week goes into CXCaller, an AI agent calling platform. The rest goes into research, and into whatever is new enough to be worth taking apart.
- 05
- Years building and exploring
- 02
- Papers at AI conferences
- 10+
- Agent and product systems shipped
Colombo, working remote
Work
What I actually do.
Three strands, and they feed each other. The research keeps the engineering honest about what these models can actually carry, and the product work keeps the research pointed at something real.
01
Agentic systems
Agents that plan, reach for tools and finish a task instead of demoing one. Built largely on Agno, with the memory, retrieval and guardrails it takes to put one in front of a customer.
02
Voice and conversation
Voice agents and chatbots that carry a real exchange end to end, and the platforms underneath them. CXCaller, the AI agent calling platform I co founded, is where most of that lands.
03
Research
Published work on cooperative multi agent systems, and a standing habit of taking apart whatever is new enough to be worth the time.
Where
Where the work happens.
A company of my own, and two engineering teams I build with.
2025 to present
Now
Articom Inc
Software Developer
Agentic AI chatbots, voice agents and agent calling tools, built largely on the Agno framework.
2024 to present
Now
Inforwaves
Associate Software Engineer
Product engineering, and the first agentic AI components the team put into production.
Research
Agents that keep talking under attack.
Two years on adversarially robust emergent communication between cooperative agents, published at two international AI conferences. The short version: agents that invent their own language will happily be lied to, and this is what it takes to stop that.
Publications
Message Masking and Curriculum Adversarial Training for Robust Emergent Communication in Cooperative Multi Agent Systems
7th World Symposium on Artificial Intelligence (WSAI 2026), Jinan, China
Adversarially Robust Emergent Communication in Cooperative Multi Agent Reinforcement Learning
ADScAI 2026
Writing
Articles.
Notes from the build, outside the conference papers.
Contact