All projects
AIProfessional

Bolka — Voice AI Platform

Built the frontend for Bolka, an enterprise voice AI platform that lets businesses deploy conversational AI agents on websites and share via links. The platform uses browser-native WebRTC, supports API streaming, and integrates Google Speech-to-Text and Text-to-Speech for real-time voice conversations.

RoleSoftware Engineer (Frontend)
TypeProfessional
Stack
React.jsNext.jsTypeScriptWebRTCWebSocketsTailwind CSSGoogle Speech APIsAPI Streaming

Problem / User need

Businesses needed a way to deploy voice AI agents on their websites without complex telephony infrastructure (SIP trunks, Twilio). Users needed to click, talk, and get answers directly in the browser — 24/7, with no app downloads.

Solution & design decisions

Implemented browser-native WebRTC for peer-to-peer audio streaming, integrated Google Speech-to-Text and Text-to-Speech APIs for the voice pipeline, and built the embeddable Bolka Embed widget as a self-contained React component. Bolka Share was built as a shareable link system that opens the voice agent in any browser without downloads.

Technical implementation

  • Built Bolka Embed — an embeddable voice AI widget deployable via a single script tag, with customisable colour scheme and agent personality.
  • Built Bolka Share — a shareable link product that opens a branded voice agent session in any modern browser.
  • Implemented WebRTC-based real-time audio capture and streaming with low-latency connection handling.
  • Integrated Google Speech-to-Text (STT) and Text-to-Speech (TTS) APIs for the full voice interaction pipeline.
  • Built the agent configuration dashboard where businesses create, customise, and deploy their voice agents.
  • Implemented API streaming for progressive rendering of AI responses, improving perceived response speed.
  • Delivered proof-of-concept builds for enterprise clients including HDFC within Agile/Scrum sprints.

Constraints

  • Real-time, low-latency voice communication entirely in the browser using WebRTC — no phone line infrastructure.
  • Streaming AI responses progressively rendered as audio and text to feel natural and instant.
  • Embeddable widget deployable on any website with one line of code, with customisable branding.
  • Support for shareable voice AI links usable via email, SMS, WhatsApp, or QR code.

Outcomes

  • Shipped a production voice AI platform used by multiple enterprise clients.
  • Enabled voice agent deployment with zero telephony infrastructure — browser-native only.
  • Embeddable widget deployable in minutes, adopted by clients across healthcare, insurance, and finance sectors.
  • Real-time WebRTC + streaming response pipeline delivered a natural, low-latency conversation experience.