Senior Backend Engineer

Location: Bengaluru, India

About the role

• We're hiring a Senior Backend Engineer to own high-throughput, real-time systems behind Convin's AI products: voicebots, AI agents, Auto QA and agent assist.

• Convin builds AI for contact centers. Our platform serves 100+ enterprise clients across BFSI, edtech and other sectors, handles 50M+ API requests a month, runs 500K+ voicebot calls a day and analyzes 500M+ conversations a year.

• You'll design systems end to end, ship them to production, and be on the hook for how they behave at scale. This is a role for an engineer who wants ownership, not tickets.

What you'll work on

• Real-time voice AI. Build and scale our Go voicebot across WebSocket and SIP, orchestrating STT → LLM → TTS with sub-second end-to-end latency at hundreds of thousands of calls a day.

• LLM systems in production. Design RAG pipelines, function calling (CRM updates, calendar booking, payment links) and multi-provider LLM orchestration with fallbacks, caching and guardrails against hallucination.

• Core platform services. Own high-volume services such as our rule engine, job scheduler and notification system, each running millions of operations a day.

• Scale and cost. Profile, tune and right-size services on Kubernetes; cut latency and cloud, LLM and TTS spend without hurting quality.

• Data layer. Model data and optimize queries across PostgreSQL, Redis, ClickHouse and Qdrant.

• Observability and reliability. Instrument services with Prometheus, Grafana and Jaeger; define SLOs, run on-call for what you own, and write blameless postmortems.

• Engineering leadership. Write design docs, review code, mentor SDE1/SDE2 engineers and interns, and raise the bar on code quality.

• Customers. Join technical discussions with enterprise clients when a design needs to fit their telephony, data or compliance constraints.

What we're looking for

• 3–5 years building backend systems, with at least 1 year owning services in production at meaningful scale.

• Strong in Go (preferred) or Python , with solid grasp of concurrency, networking and memory behavior.

• Hands-on distributed systems experience: queues (Kafka, RabbitMQ or Celery), caching, idempotency, retries, backpressure and graceful degradation.

• Production experience with Kubernetes and Docker on AWS or GCP, including autoscaling and resource tuning.

• Strong PostgreSQL skills: schema design, indexing, query plans; working knowledge of Redis.

• Has shipped at least one LLM-powered feature to real users and can talk about its latency, cost and failure modes.

• Uses metrics, logs and traces to debug, not guesswork.

• Writes clear design docs and can defend trade-offs with numbers.

• Ownership mindset: you've seen a system through launch, incidents and the rewrite after.

Nice to have

• Real-time audio or voice: WebRTC, SIP, telephony, or STT/TTS providers.

• RAG and vector databases (Qdrant or similar), embeddings, prompt engineering and LLM evaluation.

• ClickHouse, ScyllaDB or Elasticsearch for analytics or high-write workloads.

• Service mesh and infra tooling: Envoy, Istio, Terraform, Helm.

• gRPC and WebSocket services at high connection counts.

• Multi-tenant SaaS architecture, or single-to-multi-tenant migrations.

• Fluency with AI coding tools (Claude Code, Cursor, Copilot) in daily work.

• Startup experience, or having built something from zero to production.

Our stack
Area Tools
Languages Go, Python, SQL
Services REST, gRPC, WebSockets, WebRTC, SIP, Kafka, RabbitMQ, Celery
Infrastructure Kubernetes (GKE, OCI), Docker, Envoy, Istio, Terraform, Helm
Data PostgreSQL, Redis, ClickHouse, Qdrant, Elasticsearch
AI OpenAI, Anthropic, Google, Qwen; multiple STT and TTS providers
Observability Prometheus, Grafana, Jaeger, ELK

Why Convin

• Real scale, early. Your code handles millions of conversations from day one, not a demo.

• Frontier problems. Sub-second voice AI, LLM cost and latency, and agentic systems that act on customers' behalf.

• Ownership. Small team, large surface area; you own systems, not tickets.

• Growth. Our backend leads grew from within, from intern to lead.

Hiring process

• Introductory call

• LLD / machine coding round

• HLD round

• Final interview with the hiring manager