About Navana

At Navana, we are at the forefront of developing voice AI solutions tailored for Indic languages, driving growth and innovation for large enterprises. We build real-time voice AI systems that power critical customer interactions for leading enterprises. Our stack is built for low-latency, high-throughput workloads and is deployed both in the cloud and on our customer infrastructure. As we scale, security and compliance are becoming central to how we build and deploy. You will own security across our multi-cloud and on-prem footprint, embedded into how we ship and establish practices the rest of the company will build on.

Core Responsibilities

• Data Specification, QA Team & Vendor Management
  • Define Data & Quality Standards – Set detailed data requirements for training, fine-tuning, and evaluating ASR and TTS models in 10 + Indic languages; publish clear guidelines.
  • Run Internal QA Operations – Shape workflows for the in-house QA team (task assignment, spot checks, audits, feedback loops) to ensure every annotation, transcription, and evaluation meets or exceeds the defined standards.
  • Collaborate with External Vendors – Specify deliverables, develop collection processes, and enforce compliance with quality targets for outsourced data work.
  • Draft Protocols & Scopes – Write precise data-collection protocols and scope-of-work documents that cover both internal QA tasks and vendor contributions.
  • Own Timelines & Delivery – Set schedules and quality gates, track progress of both in-house and vendor pipelines, and resolve blockers to guarantee on-time delivery of high-quality speech and text corpora.
• Error-Analysis Pipeline Development
  • Design and manage large-scale error-analysis pipelines grounded in linguistic insight and deep ASR/TTS knowledge.
  • Build LLM-powered analysis tools: create prompt-engineering workflows and custom LLM utilities that automatically identify errors, surface root causes, and generate diagnostic reports.
  • Develop robust methodologies to uncover, diagnose, and categorize model errors (e.g., mis-recognitions, prosody failures, hallucinations).
  • Present error-analysis insights and recommendations to product managers, speech scientists and enterprise BFSI stakeholders.
• Automated QA & Evaluation
  • Create and refine automated quality checks, verifying annotations and automatically spotting errors, to accelerate data throughput while maintaining stringent standards.
  • Maintain evaluation datasets and benchmarking frameworks for ASR, TTS and LLMs.
  • Conduct acoustic-phonetic and conversational analyses on production data to identify patterns, anomalies, and improvement opportunities.

Qualifications & Experience

• Master’s or Ph.D. in Computational Linguistics, Linguistics or a related field, plus a strong Computer-Science foundation.
• 2+ years evaluating ASR/TTS models, building evaluation frameworks, curating data and managing vendor partnerships—preferably with Indic languages or other low-resource settings.
• Demonstrated success in building, leading and scaling QA or linguistics teams (hiring, mentoring, performance management).
• Hands-on experience creating curated data for ASR, TTS and speaker identification/diarization.
• Familiarity with finite-state transducers (FSTs) and using frameworks such as OpenFST and Pynini.
• Expertise in audio analysis using tools such as Praat / Audacity and familiarity with annotation platforms.
• Demonstrated success in defining data specifications, enforcing QA processes, and ensuring high-quality deliverables.

How to Apply