Software Development Engineer II - Multichannel Agents

Overview

An SDE II role on the Multichannel team at Blue Machines AI, building the channel adapters, unified conversation layer and media handling that let a single AI agent operate consistently across phone, WhatsApp, web chat and video.

What you'll do

• Build and own channel integrations: WhatsApp Business Platform (Cloud API), telephony (SIP/PSTN via carriers and CPaaS providers), the embeddable web chat widget, and in-app/web voice and video over WebRTC, from API design through production on-call.
• Design the unified conversation layer: a channel-agnostic session, message and identity model so the agent runtime sees one conversation regardless of channel, including channel switches (e.g., a call that continues on WhatsApp) and cross-channel context.
• Handle multimodal input and output: text, voice notes, live audio streams, video, images, documents, location, and interactive messages (buttons, lists, carousels), normalising inbound media and rendering responses to each channel's capabilities and limits.
• Build realtime plumbing: WebSockets/WebRTC for live audio and video, streaming responses to the widget, and hooks into the voice pipeline (STT, TTS, turn detection, barge-in) owned with Team Orca.
• Make delivery reliable: idempotent webhook ingestion, retries and dead-letter handling, ordering guarantees, delivery/read-status tracking, rate-limit and quality-tier management on WhatsApp, and graceful degradation when a provider is down.
• Ship the web widget as a product: lightweight embeddable SDK, theming and configuration from the Console, reconnect and offline behaviour, attachment uploads, accessibility, and security for running inside third-party pages.
• Own it in production: instrument latency, delivery success and error rates per channel; build dashboards and alerts; debug live incidents across services, carrier/BSP logs and customer networks.
• Write design docs, review code, and raise the bar on testing for asynchronous, event-driven systems. Use AI coding tools as a normal part of how you work, with no compromise on quality.

What you'll need

• 3–6 years building and running backend or full-stack systems in production, with clear ownership of at least one service end to end.
• Strong in one of Python, Go, Node.js/TypeScript or Java, and comfortable reading the others.
• Solid fundamentals in distributed, event-driven systems: queues/streams (Kafka, Redis Streams, RabbitMQ, SQS or similar), async processing, idempotency, retries, and consistency trade-offs.
• Hands-on experience integrating at least one messaging or communication channel (WhatsApp Business API, SMS, telephony/CPaaS such as Twilio, Plivo, Exotel, Vonage or similar, or a chat platform), including webhooks, signature verification and provider-specific quirks.
• Working knowledge of realtime protocols: WebSockets, and ideally WebRTC or streaming media.
• Comfortable designing APIs and data models (SQL and NoSQL) that will be used by other teams and by customers.
• Production instincts: reaches for logs, traces and metrics before guessing, and designs for failure because providers will fail.
• Clear written communication; design docs and incident notes should stand on their own.

Nice to have

• Built a chat or voice widget/SDK embedded in third-party websites or mobile apps.
• Experience with SIP, RTP, media servers (FreeSWITCH, Asterisk, LiveKit, Janus, mediasoup) or codecs and audio processing.
• Familiarity with WhatsApp specifics: template messages, 24-hour customer-service window, interactive messages, Flows, BSP integrations, quality ratings and messaging limits.
• Experience with LLM-based or conversational AI products, especially streaming responses or voice agents.
• Exposure to additional channels: RCS, Instagram/Messenger, Telegram, email, or in-app chat.
• Performance and load-testing realtime systems at high concurrency.

Details

• Location: Bengaluru, Karnataka, India.