Kian Kashfipour

B2B AI voice assistant platform for inbound/outbound call automation, sold as a white-label solution. Joined as a software engineer, built the entire AI and telephony layer from scratch, and grew into technical lead and project owner. Platform now serves +60 active clients handling +5,000 daily calls.

• Built the complete real-time voice AI pipeline, integrating VoIP providers (Twilio, Telnyx) with AI services (OpenAI, Gemini, ElevenLabs), engineering latency optimization, VAD tuning, and streaming audio handling.
• Designed a function-calling tool system using the registry pattern, enabling LLM agents to perform actions during calls: forwarding, WebRTC live transfers, Google Calendar reservations, SMS confirmations, and catalogue-based ordering.
• Built a RAG pipeline with Weaviate vector database for domain-specific question answering during live calls.
• Designed and built an outbound call center module with campaign management, expanding the platform from inbound-only to a complete call center solution.
• Implemented Temporal for orchestrating outbound workflows, call scheduling, per-step timing, and configurable retry strategies for reliable execution of long-running campaigns at scale.
• Built multi-tenant white-label architecture with isolated databases per tenant and request-level domain resolution for reseller partner onboarding under custom domains.
• Built backend with Python (FastAPI), PostgreSQL with async SQLAlchemy, Celery task queues, and CI/CD staging via Docker Swarm, Traefik, and GitHub Actions.

Ride-hailing platform with 500K+ daily rides and 20M+ users. Worked in the AI department on core data infrastructure.

• Built and maintained scalable data pipelines using Python, SQL, and PySpark for real-time processing and analytics.
• Developed a geographic search engine converting text addresses to coordinates, reducing third-party dependencies and achieving 10% reduction in search time.