Senior AI Engineer
Location: Hyderabad, Telangana, India

About the Role
GHX is building a cutting-edge LLM-powered document understanding platform focused on classification, structured data extraction, and intelligent orchestration at scale.
This is a high-impact AI engineering role where you will own the full lifecycle—from problem framing to production deployment. Initially, you will focus on prompt engineering and evaluation systems, building the quality foundation for AI performance. Over time, the role expands into agent orchestration, system architecture, and migration of rule-based systems to LLM-driven pipelines.
A strong foundation in software engineering (5+ years) is essential. This role demands engineering rigor across both traditional system design and AI system behavior.

Role Evolution
Now (0–6 months): Prompt Engineering & Evaluation
- Design and refine prompts for classification and data extraction
- Build ground truth datasets and evaluation pipelines
- Establish accuracy benchmarks and quality baselines

Mid-Term (6–12 months):



Agent Orchestration & Pipeline Design
- Develop multi-agent document processing pipelines
- Integrate MCP-based tools and external service interfaces
- Orchestrate parallel AI workflows

Long-Term (12+ months): Platform Architecture & Ownership
- Own the end-to-end document intelligence architecture
- Lead migration from rule-based to LLM-powered systems
- Define engineering standards and best practices

Core Responsibilities
1. Prompt Engineering
- Design prompts for diverse document classification and extraction tasks
- Treat prompts as formal specifications (exact, structured, and edge-case-aware)
- Develop few-shot, chain-of-thought, and structured output templates
- Manage prompt lifecycle: versioning, testing, and rollback

2. LLM Output Evaluation
- Create and maintain ground truth datasets
- Build automated evaluation pipelines (precision, recall, field-level accuracy)
- Identify and resolve conceptually incorrec

📌 Senior AI Engineer (Hyderabad)
🏢 GHX
📍 Hyderabad