Software Engineer III

Software Engineer - III

About the Role

GHX is building a next-generation Intelligent Process Automation (IPA) platform powered by LLMs and AI-native document understanding. We extract structured data from complex healthcare procurement documents — Purchase Orders, invoices, contracts — at scale across cloud environments. As a member of the IPA engineering team, you will bridge strong software engineering with applied AI. You will design and ship Python services, integrate LLM APIs, build AI agents, create skills and validate the output of LLMs to build document extraction pipelines, and own evaluation infrastructure that ensures production quality. This is not a research role — it is a hands-on engineering role where AI fluency amplifies solid software craft. 

Core Responsibilities

Python Development & Automation

• Build and maintain Python-based automation services and IPA workflows

• Develop platform-agnostic solutions supporting future migration across automation tooling

• Build reusable libraries, frameworks, and components for automation projects

• Integrate automation solutions with REST APIs and enterprise applications

• Deploy and monitor automation bots on AWS (Lambda, ECS, SQS, S3) 
  
  

  AI & LLM Integration

  • Integrate LLM APIs (like Anthropic Claude, OpenAI, Azure AI) into production pipelines
  • Design classification and extraction prompts for diverse document types (POs, invoices, contracts)
  • Write prompts that function as formal specifications — unambiguous, edge-case-aware
  • Build and iterate few-shot, chain-of-thought, and structured output templates
  • Own prompt library versioning, rollback strategy, and prompt lifecycle management 
    

    AI & LLM Integration

    • Integrate LLM APIs (like Anthropic Claude, OpenAI, Azure AI) into production pipelines
    • Design classification and extraction prompts for diverse document types (POs, invoices, contracts)
    • Write prompts that function as formal specifications — unambiguous, edge-case-aware
    • Build and iterate few-shot, chain-of-thought, and structured output templates
    • Own prompt library versioning, rollback strategy, and prompt lifecycle management

    Document Processing & IDP

    • Implement OCR/IDP workflows to extract and validate structured and unstructured data
    • Build confidence scoring frameworks for extracted fields with multi-factor validation
    • Design and curate ground truth datasets for classification and extraction tasks
    • Build automated evaluation pipelines tracking precision, recall, and field-level accuracy
    • Validate model outputs against intent — catch what is technically correct but conceptually wrong

    Software Engineering

    • Design and build production-grade APIs and services around LLM capabilities
    • Apply Clean Architecture or equivalent — design for testability and maintainability
    • Contribute to CI/CD pipelines, observability tooling, and deployment processes
    • Conduct code reviews, establish coding standards, and mentor junior engineers
    • Troubleshoot and optimize existing automation workflows and Python services
      

      Collaboration & Communication

      • Work closely with business analysts and stakeholders to understand process requirements
      • Participate in Agile ceremonies — sprint planning, retrospectives, and estimation
      • Communicate system constraints in product language; translate product needs into system boundaries
      • Surface quality metrics and model behaviour clearly to non-technical stakeholders

      Required Qualifications

      • Bachelor's degree in Computer Science, Engineering, or related field
      • 5–9 years of software engineering experience, with at least 4 years in Python
      • Strong Python skills — Pandas, NumPy, Flask/Django, async programming
      • Hands-on experience with LLM APIs (like Anthropic Claude, OpenAI, or Azure AI)
      • Prompt engineering experience — structured outputs, few-shot, chain-of-thought
      • 2+ years of experience with SQL and NoSQL databases
      • Experience deploying applications on AWS — Lambda, SQS, S3, ECS
      • Knowledge of Docker, Git, and CI/CD pipelines (Jenkins or equivalent)
      • Familiarity with REST API design and tools like Swagger and Postman
      • Understanding of OCR/IDP tools — AWS Textract, UiPath Document Understanding, or ABBYY 
        
        

        Desired Qualifications

        • Experience with AI agent orchestration frameworks — LangChain, LlamaIndex, or MCP
        • Familiarity with evaluation frameworks for LLM outputs (precision, recall, F1 at field level)
        • NLP / text classification background
        • UiPath Associate or Advanced RPA Developer certification (or equivalent)
        • Exposure to Clean Architecture, DDD, or equivalent design patterns
        • Understanding of Agile/Scrum methodologies

        What Good Looks Like

        • Prompt as specification — treats prompts like contracts: precise, unambiguous, edge-case-aware from the start
        • Verification instinct — reads AI output critically; catches the subtly wrong answer before it becomes a data quality incident
        • Evaluation-first mindset — doesn't trust a prompt in production without a test suite; builds measurement as part of the feature
        • Engineering rigour — applies the same discipline to Python services as to LLM pipelines; Clean Architecture is the default, not the exception
        • Communication range — can speak to a non-technical ops manager and a distributed system design at the same level of precision