Ahsan U Rehman

I’m an Agentic Engineer and AI Solutions Architect focused on Generative AI, multi-agent systems, and agentic workflow orchestration. I design and operate production-grade AI infrastructure—like autonomous conversational voice agents running 24/7 for client operations—and I build “self-improving” agent systems that stay reliable under real-world conditions.

In my flagship project, I architected a markdown-driven multi-agent operating system coordinating eight parallel sub-agents across deep research, outreach, content, evaluation, and reliability engineering. I delivered ~50,000 lines of TypeScript, implemented runtime observability (session logs, FIFO-pruned memory, atomic CAS locks, leak detection hooks, audit log streaming), and root caused and fixed an Anthropic Claude Code silent-write bug by introducing an inline content contract adopted across all sub-agents. I also built AI-native SaaS workflow automation for accounts payable and customs declarations, plus a LinkedIn automation platform integrated with Apify and Unipile to run high-volume B2B outreach sequences.

I’m equally strong in “trusted execution” for enterprise agents: I pioneered pre-send leak detection hooks, maintained a leak-pattern catalogue, and added a four-tier autonomy model that gates every external action for regulated contexts. Alongside that, I’ve automated operations using low-code orchestration (Make, n8n) and delivered large ROI—improving customer response times by over 80% with voice automation and reducing manual data handling by ~90% with optimized LLM prompting. Previously at PwC and in business consulting roles, I led enterprise Oracle architecture and IFRS-aligned process design, delivering high-volume performance and stakeholder leadership as a team lead of 8.