Backend Engineer AI

Atlys seeks a Backend Engineer AI to build production systems that use AI to make visa processing faster and simpler. The role is based on-site at its Delhi headquarters in New Delhi, India.

Responsibilities

• Build Python backend services that orchestrate LLMs, agents and tool calls across core product workflows.
• Create evaluations and reinforcement learning loops to measure model quality on real workflows and improve agent performance, including through reward modelling, RLHF/RLAIF and RL fine-tuning.
• Design asynchronous, event-driven workflows for peak demand while managing latency and token costs.
• Own systems from design through production, with testing, monitoring, observability and evaluations.

Ideal candidate:

• Has 4–6 years of backend engineering experience, including hands-on experience building and running LLM-powered features in production.
• Is strong in Python and async Python and has shipped production services using frameworks such as FastAPI, Django or Flask.
• Has worked with LLM APIs such as OpenAI, Anthropic or Gemini, and with prompt design, tool calling, structured outputs, RAG and vector databases such as pgvector, Pinecone or Weaviate.
• Can evaluate and harden AI systems using evaluation datasets and LLM observability tools such as Langfuse or LangSmith, while balancing accuracy, latency and cost.
• Has fundamentals in databases, distributed systems and cloud infrastructure, including PostgreSQL, Redis, message queues such as Kafka, SQS or Celery, AWS or GCP, Docker and Kubernetes.

Strong pluses include exposure to agent frameworks such as LangGraph or LlamaIndex; fine-tuning, reinforcement learning or self-hosted models; Playwright browser automation; or work in fintech, travel or other high-compliance domains.

Benefits include competitive compensation and equity, family health insurance and preventive check-ups, paid leave, office meals, learning and development credits, relocation support, and work equipment.