Founding Engineer

SoftTalk | Noida, India | Hybrid

About SoftTalk

SoftTalk is building an AI Voice Agent platform for businesses.

We’re onboarding our first few customers and are looking for a Founding Engineer to help build the product for them and shape the core platform alongside me.

This is an early-stage, fast-paced role. You’ll be close to customers, shipping quickly, solving real production problems and making core engineering decisions.

What you’ll own

You’ll work across the systems that power a real-time AI phone conversation:

Telephony → Audio Streaming → STT → Turn Detection → LLM → TTS → Audio → Telephony

You’ll help build and improve:

- Real-time voice pipelines using Pipecat
- Telephony and audio streaming using Telephony, WebSockets and async systems
- Streaming STT, LLM and TTS integrations
- Conversation orchestration and state management
- Barge-in, interruption handling and turn-taking
- Latency optimisation across the entire call pipeline
- Hold detection, grace periods and call-state handling
- Multi-provider architecture for LLM, STT and TTS
- Knowledge base and context handling for voice agents
- Tool calling, APIs and integrations with customer systems
- Backend services, APIs and platform infrastructure
- Call analytics, observability, monitoring and debugging
- Reliability, concurrency and scalability for production workloads

The kind of problems you’ll solve How do we make an AI conversation feel less like a bot?

How do we reduce end-to-end response latency?

How do we handle 100s or 1,000s of concurrent calls reliably?

How do we switch between different LLM, STT and TTS providers without rewriting the platform?

How do we handle interruptions, silence, call transfers, failures and edge cases gracefully?

How do we build the underlying infrastructure so enterprise customers can depend on it?

These are core product problems, not just implementation tasks.

Our current stack

Python, Pipecat, FastAPI,



AsyncIO, WebSockets, Telephony, STT, LLM, TTS, Docker, AWS

You do not need experience with every technology above. Strong engineering fundamentals matter more.

Who we’re looking for

We’re looking for someone who:

- Has strong experience building backend or distributed systems
- Is comfortable with Python
- Understands asynchronous and real-time systems
- Has worked with APIs, WebSockets, streaming or event-driven systems
- Is comfortable working with LLMs and up-to-date AI APIs
- Can take a problem from architecture to production
- Debugs systematically and cares about performance and reliability
- Is comfortable making technical decisions with incomplete information
- Wants ownership rather than a narrowly defined engineering role

Experience with Voice AI, Pipecat, telephony, SIP, STT/TTS, WebRTC or audio streaming is a strong plus. What this role is not

This is not a role where you’ll receive tickets and implement predefined features.

You’ll work directly with the founder on:

Architecture → Product decisions → Implementation → Deployment → Production debugging

You’ll have meaningful influence over how SoftTalk’s core platform is built.

Why join early

You’ll be joining at the stage where foundational engineering decisions still matter.

You’ll help decide how the voice stack is architected, how the platform scales, which abstractions we build and how the product evolves as we move toward enterprise customers.

There is significant ownership and ESOP participation for the right person.

Location

Noida preferred | Hybrid

Compensation

Competitive salary + ESOPs

How to apply

Send me

1. Your resume
2. GitHub / projects / demos, if available
3. A few lines about the most interesting engineering problem you’ve solved
4. Anything you’ve built involving AI, real-time systems, voice, distributed systems or high-concurrency workloads

Email: [email protected] This role is for someone who wants to build the core of the company, not just work at it.

📌 Founding Engineer (Noida)
🏢 SoftTalk
📍 Noida