I’m the Founder & CTO of Cognifyr and an engineer building open-source infrastructure that lets AI agents see, remember, orchestrate, and act without hiding authority, evidence, or provenance.
Portfolio · Cognifyr · LinkedIn · Email
| Capability | Project | What it provides |
|---|---|---|
| Govern & act | Maqam · GitHub | Policy checks, exact one-use approvals, registered execution, and verifiable receipts for agent actions. |
| See | Cockroach Crawler · GitHub | Bounded crawling, browser rendering, extraction, source routing, and normalized evidence without unrestricted browser authority. |
| Remember | Qarinah · GitHub | Local-first, evidence-linked project memory that compiles compact cited context for coding agents. Its published benchmark reports 98.71% less estimated repeated context. |
| Orchestrate | ProductLoop OS | Composable workflow, policy, approval, connector, skill, evaluation, provenance, and research packages around the stack. |
Together, these projects form a practical control layer for agents that need real tools, durable context, public-web evidence, and human authority.
Through Cognifyr, I have built and led:
- Enterprise knowledge assistants with SSO, permission-aware retrieval, citations, admin controls, and audit logs.
- Private on-premises AI systems for sensitive healthcare research.
- Real-time voice agents and workflow automation.
- Multi-tenant SaaS and internal business platforms.
Founder & CTO at Cognifyr · B.Tech in Computer Science from Cochin University of Science and Technology · Winner of the Arbitrum Open House and BNB Chain Global Hackathon.
Core stack: TypeScript, Node.js, Python, Next.js, PostgreSQL, Docker, Azure/AWS, RAG, MCP, and agent evaluation.
I’m looking for contributors, early users, and design partners working on enterprise agents, coding agents, secure tool execution, context infrastructure, or governed automation.
Start in the relevant repository’s issues or discussions, or reach me at ajnasulqubra@gmail.com.