Monoedge is looking for an AI / RAG / Applied ML Engineer to build reliable, grounded, and production-ready AI systems.
The role focuses on RAG, natural-language-to-SQL, retrieval quality, LLM evaluation, and applied machine learning, with a strong emphasis on correctness and preventing hallucinated answers.
Key Responsibilities
- Build natural-language-to-SQL systems with validation and self-correction mechanisms.
- Improve retrieval quality using vector embeddings, semantic retrieval, graph-based retrieval, and ranking.
- Ground AI-generated answers in real data and provide appropriate citations.
- Build validation mechanisms to prevent hallucinated values and unsupported claims.
- Implement query safety and strict multi-tenant data isolation.
- Improve automated analytical findings, including quality, ranking, severity calibration, and business-readable output.
- Build evaluation frameworks with correctness and safety gates to prevent regressions.
- Optimize LLM cost, latency, prompts,
and model selection.
Requirements
- Production experience with LLM / RAG systems, not just prototypes.
- Solid understanding of embeddings and vector search.
- Strong Python and SQL skills.
- Experience designing evaluations and benchmarks for LLM correctness and safety.
- Strong verification mindset and ability to validate model-generated outputs before trusting them.
Nice to Have
- Local embedding models such as sentence-transformers / BGE-class
- pgvector
- Agentic or multi-step LLM frameworks such as LangGraph
- Graph-based retrieval and knowledge graphs
- Statistics or causal-inference background