Job Description
About Us
FIH Consultancy is an operations and GTM consultancy helping international clients streamline their processes, build scalable systems, and run data-driven operations. We work with SaaS, logistics, and tech-enabled companies to design workflows, implement tools, and provide ongoing ops support.
We are now hiring a Junior Prompt Engineer for our Islamabad office (onsite/hybrid) who is comfortable working with AI prompt frameworks, evaluation tools, and structured data. This role is ideal for someone who enjoys hands-on experimentation with AI models, data validation, and building reliable workflows to extract structured information from unstructured sources such as PDFs.Role Overview
As a Junior Prompt Engineer, you will assist in writing, testing, and refining prompts designed to extract structured text from documents (primarily PDFs). Your work will involve evaluating prompt performance, managing input/output schemas (JSON), and ensuring consistent, high-quality extraction results.
This is a hands-on, analytical role focused more on prompt execution and evaluation than AI model development. You’ll collaborate closely with both technical and operations teams from our Islamabad office to ensure smooth data extraction workflows and efficient prompt testing processes.
Key Responsibilities
Prompt Development & Evaluation
- Write and refine prompts for text extraction from PDF and other document types.
- Run and analyze prompts using evaluation tools such as Vallum and BrainTrust.
- Develop and maintain evaluation sets to measure accuracy and consistency.
- Analyze errors, identify weaknesses, and iteratively improve prompts.
Schema & Data Validation
- Understand and manage JSON object schemas for input and output mapping.
- Validate extracted text against predefined schema requirements.
- Adjust prompts and logic to ensure outputs align with business rules and expected structures.
Documentation & Collaboration
- Maintain records of prompt versions, test results, and schema configurations.
- Work collaboratively with developers, data specialists, and QA to troubleshoot data or extraction issues.
- Communicate results and findings in structured formats for internal documentation.
Requirements
Experience & Knowledge
- Basic understanding of prompt engineering principles and LLM behavior.
- Familiarity with evaluation tools like Vallum, BrainTrust, or similar.
- Comfort working with structured data formats (JSON, CSV, Sheets).
- Prior experience or strong interest in text extraction workflows, PDF parsing, or LLM-based data extraction preferred.
Skills
- Excellent attention to detail and ability to identify data inconsistencies.
- Strong analytical thinking and structured problem-solving skills.
- Comfort interpreting evaluation metrics and tuning prompt parameters.
- Good written and verbal English communication.
- Willingness to work onsite/hybrid from our Islamabad office, following designated working hours aligned with project needs.
Tools
- Experience or familiarity with Vallum, BrainTrust, or similar evaluation frameworks.
- Comfortable using Google Sheets, Slack, and workflow tools (ClickUp, Notion, etc.).
- Bonus: Knowledge of Python, APIs, or data automation concepts.
Nice-to-Have
- Experience using LLMs (OpenAI, Anthropic, etc.) for structured data tasks.
- Exposure to PDF text extraction libraries (PyPDF, OCR tools, etc.).
- Understanding of prompt evaluation pipelines or automation workflows.
What We Offer
- In-house/hybrid role based in our Islamabad office, serving international clients.
- Opportunity to gain hands-on experience with cutting-edge AI and automation workflows.
- Systems-driven work culture focused on documentation, structure, and process improvement.
- Competitive salary with growth opportunities based on performance.
- Collaborative and supportive team environment.