Senior AI Parsing Engineer: LLM Extraction & Validation

Monks

Colombia

Presencial

COP 167.400.000 - 234.360.000

Jornada completa

Hace 7 días

Sé de los primeros/as/es en solicitar esta vacante

Generador de candidaturas

No envíes un currículum genérico — crea un currículum y una carta de presentación adaptados a este puesto concreto.

Supera los filtros ATS

Descripción de la vacante

Monks is seeking an AI / Parsing Engineer to own pipelines that transform messy documents into structured data for agentic AI systems. You will blend LLMs with traditional parsing, OCR, and layout analysis to deliver reliable extractions.

The role requires 5+ years shipping production software, experience with TypeScript/Node.js or Python, and strong English skills. You'll work in a fast-paced consulting environment within Monks' AI ecosystem.

Formación

5+ years of experience shipping production software.
Experience extracting structured data from unstructured sources using LLMs, NLP, and/or OCR.
Experience with LLM structured outputs, tool calling, and prompt iteration.
Experience developing in TypeScript and Node.js, or Python.
Familiarity with document AI tools and techniques, including OCR and layout parsing.
Experience building and deploying solutions on AWS.
An evaluation-driven approach to quality, validating results, and iterating based on evidence.
Strong understanding of edge cases, confidence scoring, schema validation, and nondeterministic systems.
Professional English proficiency at B2 level or higher.
Strong written and verbal communication skills, including explaining system capabilities clearly.
Comfort in a fast-paced, agile consulting environment.
Experience using AI coding agents such as Claude Code or Codex.
Experience delivering extraction systems in regulated/high-stakes industries is a plus.
Experience processing handwriting, multilingual documents, or degraded scans is a plus.

Responsabilidades

Build extraction pipelines that transform documents, transcripts, and records into structured, schema-validated data.
Combine LLMs with OCR and layout analysis for reliable extraction while controlling costs.
Design validation rules and confidence-scoring workflows with human review for low-confidence cases.
Create evaluation datasets and testing frameworks to measure extraction accuracy and regressions.
Process diverse inputs from clean PDFs to degraded scans and varied formats.
Collaborate with clients, product teams, and engineers to ensure clean data flows.
Communicate pipeline capabilities, limitations, quality thresholds, and risks clearly.
Utilize AI coding agents like Claude Code and Codex in daily workflows.