Machine Learning Engineer, Retrieval

Brilliant Systems; United States; Remote; USD 90,000 - 140,000; Full time

Benefits offered by this job

Permanent, full-time role (remote/Lah0

Job summary

Brilliant Systems is seeking a mid-to-senior professional to own retrieval quality for AI systems, shaping information models before answer generation. The role spans data prep, search, ranking, evaluation, and production monitoring to drive measurable improvements.

Responsibilities include building pipelines for unstructured client data, offline evaluation, and advancing hybrid ranking methods. Remote-friendly with a base option in Lahore, compensation discussed in first conversation.

Qualifications

• Experience with embeddings, vector search, and diagnosing retrieval failures.
• Ability to evaluate retrieval quality and link results to production outcomes.
• Experience applying ranking approaches to real-world data and choosing solutions based on evidence.

Responsibilities

• Build and tune retrieval pipelines for client documents and data not originally structured for search.
• Develop and run offline evaluations predicting production behavior.
• Improve ranking with hybrid keyword and vector search, reranking, and simpler methods where they perform better.
• Instrument retrieval in production to detect quality regressions early.
• Share findings with colleagues and occasionally contribute to technical writing.

Skills

embeddings

vector search

retrieval evaluation

production metrics

ranking methods

Job description

Role overview

Own retrieval quality for AI systems by shaping the information models receive before generating answers. The work spans document preparation, search and ranking, evaluation, and production monitoring, with the goal of making retrieval improvements measurable and reliable.

Responsibilities

• Build and tune retrieval pipelines for client documents and data that were not originally structured for search.
• Develop and run offline evaluations, checking how well they predict production behavior.
• Improve ranking with hybrid keyword and vector search, reranking, and simpler methods where they perform better.
• Instrument retrieval in production so quality regressions can be detected early.
• Share findings with colleagues and occasionally contribute to technical writing.

Requirements

• Practical experience with embeddings, vector search, and diagnosing retrieval failures.
• Ability to evaluate retrieval quality and connect measurement results to production outcomes.
• Experience applying ranking approaches to real-world data and choosing solutions based on evidence.

Benefits and work setup

• Permanent, full-time role based in Lahore or remote.
• Mid-to-senior level; compensation is discussed in the first conversation.