I

Artificial Intelligence Engineer (Irving)

InfoVision Inc.
4 hours ago
Full-time
On-site
Irving, Texas, United States
Job Role: AI Engineer Job location: Irving, TX Job type: Contract

Role Summary - AI Modernization Factory JD

AI Engineer to implement the design, development, and evolution of the client Application AI Modernization Factory—an AI-powered platform that automates and accelerates the modernization of large-scale enterprise legacy applications.

This VS Code extension-based solution leverages Large Language Models (LLMs), knowledge graphs, adaptive questioning, and code generation to transform legacy Java/Oracle enterprise systems into modern architectures such as NSA.

The role involves driving the end-to-end technical vision of the AI factory—from intelligent source code analysis to automated artifact generation—while collaborating closely with modernization teams, platform engineers, and AI specialists to continuously improve throughput, accuracy, and coverage.

Key Responsibilities

GenAI Engineering Implement the prompt engineering system — structured YAML and Markdown prompt templates with dynamic placeholder substitution, priority filtering, category routing, and multi-instance LightRAG targeting. Build and refine the Adaptive Questioning Framework — an LLM-driven recursive questioning engine with configurable probing levels, depth control, SQL indirection detection, and migration-critical validation guarantees. Implement and maintain MCP server integration with tools for vector store operations (upsert, search), Neo4j graph database queries, and file metadata lookup.

Platform Development Design, build, and maintain the VS Code extension (TypeScript/Node.js) that powers the AI Modernization Factory, including the chat participant, command handlers, and guided conversational workflows.

Design and implement the multi-step modernization pipeline: application selection → module-level targeted analysis → adaptive deep-dive questioning → LLD generation → code instructions → test instructions → implementation guidance.

Design and evolve the modular extension codebase architecture: - Services layer (LLM, session, file, user, adaptive questioning) - Handlers (chat participant, conversation, API, flows) - Utilities (embedding, token management, error tracking, SQL detection) - UI components (buttons, markdown rendering, progress indicators)

Implement a tiered error handling strategy based on analysis progress: - Early failure: stop + offer connectivity test - Mid failure: pause + auto-retry with exponential backoff - Late failure: continue with partial results

Support error classification categories: NETWORK, AUTH, SERVER, TIMEOUT, UNKNOWN.

Maintain build and packaging pipelines — bundling, TypeScript strict compilation, and automated VSIX packaging.

Integrate the VS Code extension with LightRAG server endpoints that host ingested legacy application codebases, managing connection lifecycle, endpoint targeting, and query routing to retrieve contextually relevant legacy code chunks for downstream analysis.

Collaborate with the LightRAG platform team to align on server-side ingestion pipelines, endpoint contracts, and RAG retrieval quality; partner with peer AI engineers to coordinate extension enhancements and maintain shared architectural decisions.

Python Services Maintain the Python services for performing vector operations — cosine similarity and batch cosine similarity — with JSON-based TypeScript ↔ Python subprocess interop and automatic TypeScript fallback on failure.

Manage embedding operations including external embedding API integration with batch processing, exponential backoff retry, and configurable batch sizes.

What You’ll Work On

Prompt Engineering System: YAML/Markdown prompt loader with dynamic filtering, placeholder substitution, and multi-instance routing.

AI Chat Agent: VS Code chat participant with guided modernization workflows.

Adaptive Questioning Engine: LLM-driven recursive analysis with configurable depth, probing levels, SQL indirection detection, and migration-critical enforcement.

Knowledge Graph Integration: LightRAG + Neo4j pipeline for context-aware legacy application analysis.

Artifact Generation Pipeline: Automated LLD, code instructions, and test instructions generation aligned with enterprise coding standards.

MCP Server & Tools: Vector store, graph database, and file metadata tools.

Late Chunking & Embedding: Semantic context retrieval pipeline reducing LLM token usage.

Python Vector Services: High-performance embedding and similarity operations with zero dependency.

Technical Skills Languages: TypeScript, Python, SQL Runtime: Node.js, Python Gen AI: Prompt engineering, token optimization, multi-model orchestration, RAG architectures, MCP Platform: VS Code Extension development, VS Code API, Chat Participant API, Language Model API, VSIX packaging Data Formats: YAML, Markdown, JSON