Parsing Architect (AI) - Senior
Chantilly, VA • Full-time
Posted 18d ago
Job Location
Chantilly, VA
Tech Stack
Remote Work Policy
On-site
Employment Type
Full-time
Categories
AI Engineer
About the job
Fantom Corporation is seeking a Parsing Architect (AI) to build, optimize, and maintain document extraction and chunking pipelines. These pipelines are crucial for supporting large language model (LLM) applications, vector databases, and retrieval-augmented generation (RAG) frameworks. The role involves transforming complex, unstructured documents into clean, Markdown-formatted content while preserving their original structure and context. You will collaborate with AI researchers, data engineers, and ML platform teams to enhance data quality for accurate embeddings, effective retrieval, and reliable AI responses.
Responsibilities
- Develop and maintain text extraction and preprocessing pipelines for various document types, including those with charts and tables.
- Preserve document structure, such as heading hierarchies and relationships between content sections, during extraction.
- Transform extracted content into clean, consistently formatted Markdown for embedding and AI processing.
- Implement advanced chunking strategies, like parent-child document relationships and sliding-window chunking, to maintain context and improve retrieval.
- Utilize tools like Docling and layout-aware libraries for processing complex documents.
- Implement multimodal, vision-based methods for extracting tables and interpreting document layouts.
- Write specialized parsing code to extract tabular data while preserving relationships between headers, rows, and values.
- Enrich text chunks with metadata (e.g., source lineage, temporal information, classification levels) for precise filtering and hybrid search.
- Integrate parsing outputs with vector databases, LLM applications, and RAG frameworks.
- Benchmark and tune pipelines for extraction quality, retrieval accuracy, embedding effectiveness, and context window utilization.
- Collaborate with AI researchers, data engineers, and ML platform teams to resolve parsing issues and improve pipeline performance.
Requirements
- Active Top Secret Clearance required.
- Willingness and ability to obtain a polygraph.
- Minimum of 8 years of relevant experience in software engineering, data engineering, document processing, AI data pipelines, or related technical disciplines.
- Experience developing document parsing, text extraction, and preprocessing workflows for unstructured data.
- Ability to preserve semantic structure and produce clean, structured Markdown from complex documents.
- Experience implementing chunking strategies that maintain context, including parent-child relationships and sliding windows.
- Experience with document processing tools such as Docling or comparable layout-aware libraries.
- Experience extracting tabular data from complex documents using programmatic and vision-based techniques.
- Understanding of embeddings, vector databases, LLM applications, and RAG workflows.
- Experience applying metadata enrichment to support source traceability, filtering, and hybrid search.
- Ability to evaluate pipeline quality and troubleshoot issues affecting retrieval accuracy and downstream AI responses.
- Strong collaboration and communication skills.
