Senior Backend Engineer, Vision

Overview

Senior Backend Engineer (Vision) to own the architecture and reliability of the OCR/document extraction serving harness for Sarvam’s vision models at national scale.

What you'll do

• Own the end-to-end architecture of the OCR/extraction serving harness.
• Design an accuracy harness using multi-pass extraction, ensembling, and verification.
• Architect durable, resumable document workflows using Temporal.
• Lead inference serving responsibilities including batching, GPU pool management, and routing.
• Drive latency, throughput, and cost-per-page targets through measurement and profiling.
• Build observability with tracing, per-stage metrics, SLOs, alerting, and post-incident follow-up.
• Enable multi-tenancy features like isolation, rate limiting, and fair scheduling.

What you'll need

• 5–6+ years of backend engineering experience with on-call operation of high-throughput production systems.
• Deep proficiency in Go and/or Python.
• Strong distributed systems design knowledge (queues, orchestration, idempotency, retries, timeouts, backpressure).
• Production experience with Temporal or an equivalent durable execution engine.
• Kubernetes experience in production, with GPU workload scheduling a plus.
• Demonstrated experience serving ML/LLM inference in production (batching, caching, versioning, A/B rollout, latency budgeting).
• Rigor in reliability and observability, including SLO design and incident response.

Details

• Location: Bengaluru