AI SDE 3 BE

About us:

Opportunity:

You will:

• Own low-latency, cost-efficient AI endpoints and data pipelines.
• Integrate intelligent workflows into existing services with measurable impact.
• Mentor engineers and raise engineering standards.

What you will be working on

• Make the architecture scalable for growing traffic and AI workloads.
• Build frameworks to monitor, improve, and optimize AI-backed systems.
• Improve reliability, latency, and performance of both traditional and AI services.
• Design and maintain APIs in monolith and microservices environments.
• Build event-driven systems with Kafka/RabbitMQ for high-volume pipelines.
• Implement AI components: model serving, inference/generation APIs, retrieval/RAG, embeddings, rerankers, vector stores.
• Stand up evaluation + guardrails: test sets, canaries, A/B, drift detection, content safety, fallback chains.
• Build secure storage and processing for large-scale structured/unstructured data; enforce data contracts.
• Own observability: tracing, metrics, feature flags, model/version routing, SLOs, error budgets.
• Debug production issues across services and layers; lead incident response and postmortems.
• Collaborate with AI/ML engineers and data scientists to productionize models and notebooks.
• Optimize cost/latency via caching, token budgets, autoscaling, and hardware placement.

What we are looking for

• Strong experience scaling backend/distributed systems and microservices.
• Concurrency expertise; deep understanding of reliability, performance, and resiliency.
• Event-driven architecture with Kafka/RabbitMQ; high-volume data pipelines.
• Hands-on SQL; working knowledge of NoSQL/caches; exposure to

  vector databases

  preferred.
• Production model-serving exposure: embeddings, RAG, realtime inference APIs, or eval harnesses.
• Solid in one or more of

  Go/Java/Python

  ; high-quality, maintainable code.
• Cloud deployment (AWS preferred; GCP/Azure ok). Containers, CI/CD, infra as code.
• Security and privacy fundamentals for PII and prompt/content safety.
• Nice to have: Triton/TorchServe/vLLM, quantization, OpenTelemetry, pgvector/OpenSearch/Pinecone, feature-flag platforms.
• Prior collaboration with AI/ML teams from research to production is desirable.
• Track record of scaling systems for 5+ years; ownership of production services.