Frontend Engineer
Company: Blessing Softtech
Location: Pune, Maharashtra (on-site preferred; exceptional remote candidates considered) Type: Full-time Level: Mid to Senior (3–7 years, but we hire on demonstrated ability, not years)
Why this role exists
Most AI companies ship a chat box. We ship a voice.
A user should be able to land on our site, click a button, and within 400 milliseconds be having a real conversation with an AI agent — interrupting it mid-sentence, switching from English to Hindi to Marathi, and watching the interface respond to their voice in real time.
That experience does not exist yet. Almost nobody in the world has built it well. We want the person who will.
This is not a "convert Figma to React" role. This is a role for someone who believes the browser is an underused instrument, who has opinions about audio latency, and who gets genuinely annoyed when a product feels 200ms slower than it should.
What you'll actually build
- Live voice trials on the marketing site — a visitor talks to a Fonazo agent in-browser, no signup, no download. Sub-second time-to-first-audio, graceful barge-in, visible transcription streaming word by word.
- Real-time audio visualisation — waveforms, spectrograms, agent "thinking" states, speaker diarisation cues. The interface should make an invisible technology feel physical.
- The agent builder canvas — a no-code, drag-and-drop conversation flow editor that non-technical business owners in Nashik or Surat can use without a manual.
- Cinematic product surfaces — WebGL/Three.js scenes, scroll-driven narrative, motion systems. Video-grade feel, but interactive and under 200KB of critical JS.
- Live call dashboards — thousands of concurrent conversations rendered as streaming data without dropping a frame.
- A design system that scales across four products and holds visual coherence as we grow.
What we need you to know
Non-negotiable
- Deep React (hooks, concurrency, reconciliation — you know why it re-rendered) and Next.js in production
- TypeScript, seriously used — not any sprinkled over JavaScript
- WebRTC and WebSockets — you have shipped a real-time media feature, not just read about one
- Web Audio API — AudioContext, AudioWorklet, MediaStream, buffer scheduling, echo cancellation, resampling
- Performance engineering: Core Web Vitals, bundle budgets, memory profiling, 60fps under load
- Cross-browser and mobile-web reality (Safari's audio autoplay policy should make you sigh knowingly)
Strongly desired
- WebGL / Three.js / React Three Fiber, GSAP or Framer Motion, Canvas 2D
- Streaming UI patterns: SSE, chunked rendering, optimistic state, partial transcripts
- WASM for client-side audio processing
- Voice-stack familiarity: LiveKit, Deepgram, Cartesia, ElevenLabs, Twilio Voice, Sarvam
- Accessibility and i18n across Indic scripts — Devanagari, Tamil, Bengali, Gujarati rendering is a real problem, not an afterthought
What makes us stop and read your application
- A portfolio of things you built because you wanted to see if you could
- Open-source contributions to audio, media, or rendering libraries
- An Awwwards / CSS Design Awards mention, a viral demo, a shader you're proud of
- You can argue for a design decision and change your mind when the argument is better
The kind of person we're looking for
We have been describing this internally as looking for a "Steve Jobs of India" hire — and we mean something specific by it.
Not a founder-personality. We mean taste married to engineering. Someone who will not ship a rounded corner that is 2px wrong. Someone who will rewrite a working component because the animation curve felt mechanical. Someone who understands that in a voice product, the silence between the user finishing their sentence and the agent replying is the entire product experience — and will fight for every millisecond of it.
You should be the sort of person who is a little embarrassed by work you shipped a year ago. That's the signal we're hiring for.
What you get
- Ownership. You are the frontend function. You set the standards, the stack, and the bar for everyone who joins after you.
- Real users. Fonazo runs live customer conversations for businesses in production. Nothing you build sits in a demo folder.
- Speed. We are a linear team. There is no committee between your idea and production.
- Compensation. Competitive cash, meaningful ESOP. We would rather pay for one exceptional engineer than three adequate ones.
- Access. You work directly with a globally deployed team.
How to apply
Send to [email protected] with subject line "Frontend — [Your Name]":
- Your portfolio or GitHub — lead with the single thing you're most proud of and one paragraph on why it was hard
- A short note (150 words max) on this question: A user finishes speaking. What should happen in the interface in the next 300 milliseconds, and why?
- Your CV
No cover letter templates. We read every application ourselves.
Process: Portfolio review → 45-min technical conversation → paid take-home (real problem, 6–8 hours, we pay for your time) → founder conversation → offer. Two weeks end to end.
Blessing Softtech (OPC) Private Limited · CIN: U62099PN2024OPC235937 · Pune, India
Pay: From ₹100,000.00 per year
Benefits:
- Commuter assistance
- Flexible schedule
Work Location: Remote
