Abdullrahman
HomeAboutProjectsResearchSkillsContact
Download CV
Back to Archive
Project 01 // Core Engine

LLM Auditing
Assistant.

Fine-tuned language model workflows for policy-aware document review

DOCUMENTATION
LLMFine-tuningNLPAudit Automation
LLM
LLM
FIN
Fine-tuning
NLP
NLP
AUD
Audit Automation
30%Review acceleration
20%Workflow efficiency
AI TeamLead role
Process // 01

The Challenge

Audit and compliance teams often need to review large document sets against domain-specific conditions while preserving consistency and traceability.

Bottleneck Analysis

Identification of relay failures in distributed nodes.

Process // 02

The Solution

  • Designed task-specific LLM workflows for structured review and document interpretation.
  • Combined prompt design, fine-tuning strategy, and retrieval context to improve reliability.

Advanced Core Logic

A public-safe portfolio version of fine-tuned LLM work for auditing and structured review tasks.

Fine-tuned LLM workflow
Policy-aware prompts
Evaluation loops
Traceable outputs
Architecture

Visual Chronicles

Abdullrahman

© 2026 Abdullrahman. Built for high-end AI research.

LinkedInGitHubPrivacyTerms
  • Kept the architecture public-safe by separating reusable AI patterns from confidential organizational rules.
    • Solution visualization
    audit_pipeline.py
    def review_document(document, policy_context):
    retrieved = retrieve_relevant_controls(document, policy_context)
    return llm_review(document=document, controls=retrieved, output_schema='audit_findings')