Applied AI Engineer (Document Intelligence & Prompting)

Location: United States / Remote

Company

hirify.global’s AI-first startup Hooh turns advanced AI models into practical products that solve real-world problems.

What you will do

• Analyze document layouts, sections, entities, and relations; design taxonomies and metadata schemas.
• Create and maintain prompts and chains for classification, extraction, summarization, and question answering with structured JSON outputs.
• Build hybrid retrieval using embeddings, keywords, and graph traversal, including indexing, deduplication, and drift management.
• Design model cascades, mixture-of-experts routing, and multi-step agentic pipelines with caching, batching, streaming, and retries.
• Implement result fusion, reranking, confidence checks, guardrails, and safety rules.
• Define evaluations and monitoring for quality, cost, and latency across local and hosted models.

Requirements

• Strong programming skills in Python or JavaScript/TypeScript with clear, tested, maintainable code.
• End-to-end experience shipping an LLM-powered feature from prototype to production, or equivalent open-source or portfolio work.
• Hands-on experience with RAG, vector embeddings, offline and online evaluation, A/B testing, and error analysis.
• Familiarity with model cascading and mixture-of-experts concepts, or the ability to learn them quickly.
• Practical database skills, including SQL and graph or vector database systems.
• Product mindset with a focus on measuring impact and iterating.

Culture & Benefits

• AI-first startup environment focused on turning research into usable products.
• Full-time remote work for candidates based in the United States.
• Ownership and impact in building production AI systems.
• Work with local and hosted model runtimes, vector databases, graph databases, and relational databases.