Senior AI engineer

Senior AI Engineer – Production LLM Systems
AllysAI · India · Full-time · Remote
About AllysAI
AllysAI rebuilds enterprise operations as AI-native systems. We ship production AI like products, on our AllysOS platform, for organisations in regulated industries: banking, pharmaceuticals, government and energy. We’re headquartered in Abu Dhabi, with offices in New York and Lausanne.
The role
You’ll own AI systems end to end, from design to production, and lead the engineers who build them with you. This is a hands-on role, and most of your time goes into building and shipping. You’ll also work directly with client technical teams, and you’ll need to explain and defend your design in the room.
What you’ll do
Build production RAG: chunking, retrieval, re-ranking, citations, and answers grounded only in approved sources
Build agents and tool calling, including MCP
Design evaluation and guardrails: test sets, hallucination checks, red-teaming and safe handling of regulated content
Build document AI pipelines: OCR, extraction and classification
Run LLMs in production: model choice, monitoring, cost, latency and reliability
Work with sensitive banking, pharma and government data under strict security and data-protection rules
Review code, set engineering standards and guide the team
What you bring
5+ years of software engineering, including at least 2 years building LLM or generative AI systems that run in production with real users
Strong Python (FastAPI or similar), and comfort across APIs, data and deployment
Hands-on experience with vector search and re-ranking (for example Azure AI Search, pgvector or Qdrant)
Azure experience (Azure OpenAI, AI Search, Document Intelligence, API Management) is a strong plus; AWS or GCP is also welcome
Docker, CI/CD and production monitoring
Experience in regulated environments such as banking, healthcare or government
Clear, confident English with clients;
Based in the India
What we offer
Competitive salary based on experience
Flexible time off
Real ownership: your systems go live at enterprise clients, not in demos

Work arrangement
Yes