Voice AI Research Engineer

ApplyMint; Bengaluru; On-site; INR 1,500,000 - 2,500,000; Full time

Job summary

Invorto is building cutting-edge voice AI systems powering real-world customer interactions. We are seeking a Voice AI Research Engineer to design evaluation frameworks, optimize STT‑LLM‑TTS pipelines, and ensure production-grade quality across accents and languages.

You’ll design automated testing, benchmark components, and work hands-on with STT, TTS, and ASR models, collaborating with engineering and customers to translate needs into measurable quality criteria.

Qualifications

• 4–6 years of experience in voice AI and automated quality evaluation.
• Hands-on experience with STT, TTS, and ASR models.
• Experience designing automated testing/evaluation frameworks for voice systems.
• Strong understanding of factors driving voice AI quality: accuracy, latency, robustness.

Responsibilities

• Design and build automated testing and quality frameworks for STT‑LLM‑TTS pipeline.
• Evaluate and benchmark STT/LLM/TTS components across accuracy, latency, naturalness, robustness.
• Work hands-on with STT, TTS, and ASR models for production use cases.
• Identify failure modes and edge cases and build systems to catch them before production.
• Collaborate with engineering to integrate quality checks into the development lifecycle.
• Stay current with advances in voice AI and bring new techniques to improve performance.
• Engage with customers to understand real-world use cases and translate into evaluation criteria.

Skills

Voice AI systems

Automated testing

Python

Communication skills

Research mindset

STT/ASR/TTS knowledge

Tools

Pipecat

Job description

About Us

Invorto is our Voice AI product, bringing intelligent voice agents to real-world customer and operational use cases. Our voice pipeline is built in Python, running an STT - LLM - TTS architecture on top of the Pipecat framework.

This is a chance to work on hard problems in voice AI - latency, accuracy, naturalness, and reliability - building zero-to-one, owning your area end-to-end, and shipping to production at scale.

Note: This is a customer-facing role, and strong communication skills are essential.

About the Role

We’re looking for a Voice AI Research Engineer to join the Invorto team and help build and continuously improve the voice AI systems that power our intelligent voice agents. This role is focused on the specialized craft of voice AI - designing evaluation and automation frameworks that ensure our STT, LLM, and TTS pipeline performs reliably in real-world, production conditions.

What You’ll Do

• Design and build automated testing and quality frameworks for our STT - LLM - TTS voice pipeline, built on Pipecat


• Evaluate and benchmark STT, LLM, and TTS/ASR components on accuracy, latency, naturalness, and robustness across accents, languages, and real-world audio conditions


• Work hands-on with STT, TTS, and ASR models - fine-tuning, evaluating, and improving them for production use cases


• Identify failure modes and edge cases across the pipeline (background noise, accents, interruptions, turn-taking, latency, pipeline-stage handoffs) and build systems to catch them before production


• Collaborate closely with engineering to integrate quality checks and automation into the voice agent development lifecycle within the Pipecat-based architecture


• Research and stay current with advances in voice AI, and bring in new techniques, models, and tools to improve pipeline performance


• Work directly with customers to understand real-world voice use cases and translate them into evaluation criteria and quality benchmarks


• Partner with product and engineering to define what "production-grade quality" means for voice agents and drive the team toward it

What We’re Looking For

• 4-6 years of experience, with a specialization in voice AI systems and automated quality evaluation


• Hands-on experience with STT (Speech-to-Text), TTS (Text-to-Speech), and ASR (Automatic Speech Recognition) models


• Experience designing and building automated testing/evaluation frameworks for voice or speech systems


• Strong understanding of what drives voice AI quality - accuracy, latency, naturalness, and robustness to real-world variability


• Strong programming skills in Python; familiarity with Pipecat or similar voice pipeline/orchestration frameworks is a plus


• Understanding of STT - LLM - TTS pipeline architectures and the trade-offs involved at each stage


• Research mindset - comfortable exploring new models, techniques, and tools and translating them into practical improvements


• Excellent communication skills - this is a customer-facing role, and you’ll regularly engage directly with customers to understand needs and validate quality expectations