Join Sarvam, a leading full-stack AI company, and engineer AI agents that power India's largest institutions. We bridge the gap between functional AI and systems relied upon by millions. Your work will involve building production-ready agents, treating them as software artifacts with versioning, rigorous testing, and clear accountability.

We are focused on building sovereign AI infrastructure and applications tailored for India. With a foundation in foundational models for Indian languages and enterprise AI platforms, we are creating the AI layer for hundreds of millions of people. Be part of a team backed by top investors, working with national and state governments, large enterprises, and developers.

Develop and deploy production-grade AI agents using the Sarvam stack, encompassing prompts, system design, workflows, and evaluation frameworks.

Strategize and map comprehensive agent coverage, meticulously defining scenarios, handling protocols, and edge cases to ensure robust performance.

Build compelling proof-of-concept demonstrations for clients, delivering convincing results rapidly.

Integrate AI agents into diverse existing systems, including core banking, CRM, ticketing, and telephony platforms.

Champion end-to-end quality assurance, defining success metrics, implementing evaluation pipelines, and driving continuous improvement.

Apply software engineering principles to AI agent development, emphasizing version control, code reviews, and regression testing for consistent quality.

We are seeking engineers with a minimum of 3 years of experience, prioritizing a proven track record of building and maintaining real-world AI systems over tenure.

Demonstrate extensive experience with Large Language Models (LLMs) and agentic frameworks, including context engineering, tool utilization, structured outputs, and memory management, with a deep understanding of large-scale failure modes.

Possess strong proficiency in Python and a broad skill set across backend services, data pipelines, APIs, and cloud infrastructure, with the ability to quickly adapt to new technology stacks.

Exhibit a strong engineering discipline, embracing version control, comprehensive testing, and reproducibility in AI development.

Be adept at defining meaningful metrics, building evaluation pipelines, and discerning genuine performance signals from noise.