Head of AI Engineering

About the role

THOUT.AI is seeking a Head of AI Engineering to define and drive the technical direction of its core AI platform. This senior, hands‑on leadership role combines architecture design, code contribution, and team building to deliver end‑to‑end LLM pipelines, agentic retrieval‑augmented generation, and scalable model serving.

Key responsibilities

• Set the technical strategy for multi‑pass LLM pipelines, transcript processing, TLDR generation, and PII detection.
• Design and maintain agentic, tool‑using LLM systems with production‑grade guardrails, evaluation pipelines, and observability.
• Make build‑vs‑buy and model‑selection decisions, balancing cost, latency, and accuracy.
• Hands‑on development and debugging of LLM pipelines, prompt‑cache management, and schema injection handling.
• Build hybrid retrieval systems (BM25, dense vectors, re‑ranking, query rewriting) optimized for meeting transcript data.
• Architect memory‑management and cross‑session context layers for long‑context conversations.
• Lead model serving and inference optimization using vLLM/TGI stacks and ensemble strategies.
• Hire, mentor, and grow a founding AI engineering team, establishing engineering standards and processes.

Required profile

• Senior technical leader with a proven track record in AI/ML system design and production deployment.
• Experience building and scaling LLM‑based products in fast‑moving environments.
• Strong problem‑solving mindset, able to shift priorities weekly while maintaining delivery momentum.
• Excellent communication skills to collaborate with founders and product teams.

Required skills

• LLM pipeline development
• Agentic RAG architecture
• BM25 and dense retrieval
• Model serving (vLLM, TGI)
• Fine‑tuning and domain‑specific extraction
• MLOps practices
• Prompt engineering and evaluation frameworks

What we offer

• Competitive monthly salary (SGD 15,000‑20,000)
• Leadership role in a high‑growth AI startup
• Opportunity to shape core product architecture from the ground up
• Equity participation and a collaborative founder‑level environment