SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Atlys is building an AI-first platform to remove friction from travel and visa processing. The company has already processed visas for 1M+ travellers across 127 countries and is expanding rapidly.
As a Backend Engineer (AI), you will build production systems that integrate large language models, AI agents, and tool orchestration into core product workflows. Your responsibilities include:
- Designing and implementing AI-native backend services in Python that orchestrate LLMs, agents, and tool calls across visa processing and travel workflows.
- Building evaluation frameworks and reinforcement learning loops to measure model quality on real-world workflows, using techniques like reward modelling, RLHF/RLAIF, and RL fine-tuning to continuously improve agent performance.
- Architecting async, event-driven workflows that scale for high volume and cost efficiency, managing peak demand while optimizing latency and token spend.
- Owning systems end-to-end from design through production, including comprehensive testing, monitoring, observability, and evaluation infrastructure.
You will work in a fast-paced, high-bar environment where deliberate execution and reliability are as valued as shipping new features. The role offers exposure to applied AI at scale in a high-compliance domain (travel/visa processing).
REQUIREMENTS:
- 4–7 years of backend engineering experience with hands-on production work building and running LLM-powered features.
- Exceptional proficiency in Python, including async Python, with shipped production services using FastAPI, Django, or Flask.
- Hands-on experience with LLM APIs (OpenAI, Anthropic, Gemini) and surrounding building blocks: prompt design, tool calling, structured outputs, RAG, and vector databases (pgvector, Pinecone, Weaviate, or similar).
- Demonstrated ability to evaluate and harden AI systems: eval datasets, LLM observability tools (Langfuse, LangSmith, or similar), and trade-offs between accuracy, latency, and cost.
- Solid fundamentals in databases, distributed systems, and cloud infrastructure: PostgreSQL, Redis, message queues (Kafka, SQS, Celery), AWS or GCP, Docker, and Kubernetes.
- Exposure to agent frameworks (LangGraph, LlamaIndex, or similar), fine-tuning, reinforcement learning (RLHF, reward modelling), self-hosted models, browser automation (Playwright), or fintech/travel/high-compliance domain experience is a strong plus.
