We're looking for a Backend Intern to join Sarvam's engineering team and own meaningful workstreams in two critical areas: our inference pipeline infrastructure and diagnostics API services. You'll work on the systems that serve AI models at scale, help build robust APIs for diagnostics and observability, and contribute to the data pipelines that keep everything running reliably. Strong performers will be fast-tracked to a full-time offer at the end of the internship. Preferred background: AI/ML or Computer Science.
What You'll Do
• Build and optimise backend services for LLM inference pipelines in Python or Node.js
• Develop and maintain diagnostics API services for model observability and health monitoring
• Integrate LLM APIs and manage request routing, latency, and error handling across inference flows
• Design and query SQL and NoSQL databases to support pipeline state management and diagnostics data
• Build and maintain data pipelines to support inference workloads and operational metrics
• Deploy and manage services on cloud infrastructure (AWS or GCP) using version-controlled codebases on Git
• Collaborate with ML engineers and platform teams to debug, profile, and improve system performance
What We're Looking For
• Proficiency in Python or Node.js; comfortable writing clean, production-quality backend code
• Solid understanding of REST API design, including diagnostics and observability endpoints
• Familiarity with SQL and at least one NoSQL database (e.g. MongoDB, Redis, or DynamoDB)
• Working knowledge of Git for version control and collaborative development
• Basic exposure to cloud platforms — AWS or GCP
• Interest in LLMs and familiarity with LLM API integration patterns
• Background in Computer Science, AI, or Machine Learning preferred
Bonus Points
• Prior exposure to inference serving frameworks (e.g. vLLM, TGI, Triton, or similar)
• Experience with monitoring and observability tooling (e.g. Prometheus, Grafana, or OpenTelemetry)
• Familiarity with containerisation and orchestration (Docker, Kubernetes)
• Contributions to open-source projects in backend or ML infrastructure