SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sarvam AI is building India's full-stack sovereign AI platform, focusing on making AI genuinely work for Indian enterprises and public institutions. The company is backed by Lightspeed, Peak XV, and Khosla Ventures, and partners with leading Indian brands including Tata Capital, SBI Life, CRED, IDFC, and LIC.
The Vision team builds vision-language models for OCR and structured extraction, handling the full complexity of Indian documents at population scale: PAN and Aadhaar cards, bank statements, GST filings, insurance and medical reports, contracts, and legal documents across multiple languages, scan qualities, and layouts.
As a Backend Engineer, you will build and own services within the document intelligence harness—the layer between vision models and enterprise consumers. You'll work on the complete pipeline: ingestion, page-level fan-out, model inference, post-processing, validation, and assembly. This is a build role with real surface area and close mentorship, designed to help you grow into owning services outright.
Key responsibilities include: building and maintaining REST APIs for document submission, job status, and result retrieval (both synchronous and async flows); implementing Temporal workflows and activities for multi-stage document pipelines; writing pre- and post-processing logic for page segmentation, de-skewing, layout handling, and schema validation; instrumenting systems to track latency and cost per stage; building internal tooling for replay, evaluation, and regression testing; debugging production issues across the stack; and collaborating directly with the models team.
The tech stack includes Go, Python, Temporal, REST, Kubernetes, PostgreSQL, Redis, object storage, and OpenTelemetry-based observability.
You should have 1–2 years building backend services in production, strong fundamentals in Go or Python, solid grasp of HTTP and REST API design, comfort with async/background job processing, working knowledge of PostgreSQL and Redis, familiarity with Docker and Kubernetes, and the ability to debug using logs, traces, and code. A genuine interest in AI systems engineering and making models work in production is essential. Bonus experience includes Temporal or other durable workflow engines, OCR/computer vision work, LLM/VLM inference, GPU serving stacks, or open-source contributions.
