SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
MaintainX is an AI-powered maintenance and asset management platform serving 14,000+ customers including Duracell, Shell, Cintas, and Brenntag. The company raised $150M in Series D funding (total $254M) and was named to the Forbes 2025 Cloud 100 list. MaintainX manages 13.9M+ assets, 79.5M+ completed work orders, and serves 150,000+ technicians generating operating data weekly.
You'll own the Generation half of Document Intelligence, a horizontal engine that transforms raw customer files into structured, trustworthy maintenance knowledge. This role focuses on turning multimodal primitives (keyframes, transcripts, OCR) into schema-valid entities like standard operating procedures (SOPs).
Key responsibilities include: designing and iterating recipe prompts (system, few-shot, and context assembly) for each generation recipe; defining per-entity target schemas and domain validators that generated output must pass; building generation-quality eval datasets and rubrics (offline and online) using LLMX's eval pipeline; assembling multimodal context windows from keyframes, transcripts, and OCR to optimize each model call; and choosing the appropriate model and token budget per recipe based on quality, cost, and latency tradeoffs.
You'll work with LLMX for model access and Attachments for ingestion, handing off validated entities to domain-owning teams. You'll report to the Engineering Lead and collaborate closely with the Processing side of Document Intelligence.
Required qualifications: strong applied GenAI craft including prompt engineering, structured output/tool-use, and RAG patterns; real eval discipline with experience building datasets, rubrics, and measuring factuality/relevance/quality; shipped LLM features into production services (not notebooks) with ability to connect model performance to product impact; comfort with multimodal inputs (video, PDF, audio, image) converted to text or structured output.
Nice-to-have skills: LLM observability/cost awareness or eval platform experience; document, PDF, or video understanding; OCR or retrieval systems experience; light fine-tuning experience or familiarity with industrial/maintenance domains.
