Nahid H.

I build AI voice agents that actually work on real phone lines , not demos that fall apart when SIP trunks, call transfers, or real carrier routing show up. 14 years in telecom and software engineering. Here is the problem I keep solving: anyone can spin up a Vapi or Retell AI agent in a few hours. But making that agent handle live calls on an Asterisk or FreePBX system — barge-in, warm transfers, outbound dialing across international SIP trunks — that takes someone who understands both the AI and the phone infrastructure underneath it. That is what I do. I work across the full stack. On the AI side: Vapi, Retell AI, OpenAI Realtime API, LiveKit, Deepgram, ElevenLabs, Cartesia. I build RAG pipelines so your voice agent pulls live answers from your CRM or knowledge base during the call, not from a canned script. On the telephony side: Asterisk, FreePBX, 3CX, Twilio, Kamailio, SIP trunk provisioning, codec handling, call recording, IVR logic. I connect the two through n8n, Make, Python, Node.js, and webhook-based automation that syncs everything back to GoHighLevel, HubSpot, or whatever CRM you run. The infrastructure runs on AWS, GCP, Docker, Kubernetes, Terraform, and Cloudflare Tunnels. I have also deployed MCP (Model Context Protocol) servers and local LLM setups for clients who need on-prem AI with full data control. Track record: voice agents that cut support tickets by 40%. Onboarding automation that eliminated 90% of manual work. 99.99% network uptime. Call-setup latency down 30%. Projects I take on: AI receptionists, outbound AI dialers, IVR-to-conversational-AI migration, custom PBX deployments with AI-powered routing, and full voice agent builds on Vapi, Retell, or custom stacks — architecture through production. Send me the scope. A short call is usually enough to map your setup and lay out exactly what the build looks like.