Overview
Product Manager for Sarvam’s Model APIs, owning how developers and enterprises consume the models while feeding learnings back into what gets trained next.
What you'll do
- Own the product surface for model APIs across speech, language, and multimodal, including specs, versioning, documentation, developer experience, and deprecation.
- Shape model roadmaps with research leads using evidence from usage, evals, and customer signals.
- Define quality criteria per model line and set targets for accuracy, latency, cost per call, robustness, and key failure modes.
- Own evaluation as a product, including benchmark suites, offline/online tests, human evals, and reporting.
- Translate customer and developer feedback into model and API requirements and communicate what changed and why.
- Partner with infra and platform teams on rollout, capacity, monitoring, and reliability in production.
- Drive end-to-end release planning, including positioning, launch, and migration paths.
What you'll need
- 3+ years building AI/ML products, developer platforms, or APIs (where the underlying model informed product decisions).
- Working fluency across the model lifecycle: data, training, fine-tuning, evaluation, deployment, and monitoring.
- Strong analytical instincts and comfort working with data.
- Sound judgment on trade-offs between quality, latency, and cost.
- Ability to work closely with researchers and earn their trust.
Nice to have
- Experience with speech (ASR, TTS) or LLM APIs and related metrics (WER, CER, MOS, TTFT, tokens per second).
- Track record shipping and maintaining a public or enterprise-facing API.
- Hands-on with SQL, Python, or ML tooling for analysis.
- Exposure to Indian language AI, multilingual systems, or evaluation for low-resource languages.
Details
Read the full description and apply on the company’s own careers page.