Senior AI Engineer — Hybrid Retrieval & Evaluation

GlobalLogic Inc.; Polska; On-site; PLN 240,000 - 360,000; Full time

Job summary

GlobalLogic seeks a skilled engineer to help design an enterprise-scale HITL AI agent platform. You will build hybrid retrieval over enterprise sources with vector and keyword search, and implement performant report generation with source attribution.

You'll work on evaluating LLMs, defining golden datasets, and driving automated checks with few-shot strategies. The role emphasizes production readiness, code quality, and collaboration across teams to deliver scalable intelligence solutions.

Qualifications

• 5+ years of engineering experience, including 2+ years on production RAG or search systems.
• Embeddings, vector databases (Vertex AI Vector Search, pgvector or similar), hybrid search and rerankers.
• LLM evaluation methods and tooling (Ragas, DeepEval, custom judges, golden-set design).
• Strong Python, from prototype through to production code.
• An information retrieval or NLP research background.
• Vertex AI Search, BigQuery, and in-context learning optimization (e.g., DSPy).
• Structured report generation and citation grounding.

Responsibilities

• Build hybrid retrieval over enterprise sources (vector and keyword search), with reranking and permission-aware filtering.
• Build the precedent store: index decision points with human decisions and rewards, and retrieve the right examples at runtime.
• Deliver report generation and explanation layers with source attribution.
• Build automated evaluation: golden datasets, LLM-as-judge, regression suites, and A/B tests of few-shot strategies.
• Apply the dynamic few-shot approach to the LLM stage of alert triage.

Skills

RAG / search systems

Python

Information retrieval

NLP research

Embeddings

Vector databases

Vertex AI Search

BigQuery

In-context learning

LLM evaluation

Golden datasets

Tools

Vertex AI Vector Search

pgvector

Job description

GlobalLogic seeks a skilled engineer to help design an enterprise-scale HITL AI agent platform. You will build hybrid retrieval over enterprise sources with vector and keyword search, and implement performant report generation with source attribution.

You'll work on evaluating LLMs, defining golden datasets, and driving automated checks with few-shot strategies. The role emphasizes production readiness, code quality, and collaboration across teams to deliver scalable intelligence solutions.