Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a custom ai voice agent with livekit or pipecat
Mahimai
About this gig
I build custom, production-grade voice agents in Python/TypeScript on LiveKit Agents or Pipecat, the open-source stacks behind today's best AI products. You own the code. No platform lock-in, no per-minute middleman markup.
WHAT YOU GET
- Full STT LLM TTS pipeline (Deepgram, OpenAI, ElevenLabs, Cartesia your pick)
- Natural turn-taking and interruption handling, tuned for sub-second latency
- Integrations: CRM, calendar booking, webhooks, your internal APIs
- Deployment to your infrastructure + docs and a video walkthrough
WHY ME
- Major contributor to livekit-agents, fastrtc (Hugging Face).
- Have built 5+ production voice solutions in Americas and European countries
- Author of the Voice Agents Handbook
- Maintainer of open-source AI tooling (envoic, OpenRTC, VoiceGateway, Voice-Prices)
- AI consulting is my full-time work, I publish on ai agent daily
THIS GIG IS FOR YOU IF
- You've outgrown Vapi/Retell pricing or hit their customization ceiling
- You need behavior no-code platforms can't do (custom turn logic, multi-agent, RAG)
- You want infrastructure you control as call volume
Get to know Mahimai
Mahimai
AI, Voice and Chatbot developer
- FromCanada
- Member sinceSep 2021
- Avg. response time1 hour
- Last delivery7 months
Languages
English, French
Hello, I am Mahimai, I have experience in: (1) Vapi Voice Agent and custom Twilio Voice,
AI/ML/DL Projects (2) Python application (FastAPI) (3) Azure AI Services Freelance experience of 6yrs.
Other AI Development Services I Offer
FAQ
Why custom code instead of Vapi or Retell?
Those platforms are great until you need control: custom turn-taking, sub-second latency tuning, multi-agent handoff, or costs that don't scale with per-minute markups. With LiveKit/Pipecat you pay raw provider costs and own everything.
Which STT/LLM/TTS providers do you support?
Any with an API. Common picks: Deepgram or AssemblyAI for STT; OpenAI, Anthropic, or Gemini for the LLM; ElevenLabs, Cartesia, or Rime for TTS. I'll recommend a stack for your latency/cost/quality priorities.
Do I own the code?
Fully. Delivered via Git repo with docs. No licenses, no lock-in to me.
What are the ongoing running costs?
Typically $0.03–$0.10/min in raw provider costs depending on the stack, far below managed-platform pricing at volume. I include a cost model with every delivery.
Can you connect it to a real phone number?
Yes, Standard and Premium include telephony (Twilio/Telnyx/SIP). For deep PBX/SIP-trunk integration, see my dedicated AI phone agent gig.
What do you need from me to start?
Your use case, example conversations or a call script, API access to systems we integrate, and cloud + provider accounts (I can set these up with you on a kickoff call).
