Real-Time AI/ML Engineer – Deploy LLMs & RAG

LockedIn AI; United States; Remote; USD 130,000 - 210,000; Full time

Job summary

LockedIn AI is seeking a full-stack AI engineer to design, train, and deploy language and speech models for a real-time AI assistant. You will build end-to-end ML pipelines, from data prep to production monitoring, bridging research and production code.

The role requires hands-on ML with transformer models, optimization, and observability, ensuring rapid, robust shipping of features that impact users directly.

Qualifications

• Strong software engineering fundamentals alongside deep machine learning expertise.
• Hands-on experience designing transformer-based architectures and taking them to production.
• Familiarity with inference optimization, deployment patterns, and production observability.
• Ability to balance model quality, user impact, and shipping speed.
• Solid understanding of evaluation, testing, and monitoring practices for ML systems.

Responsibilities

• Design, train, and fine-tune language models and speech models for production use cases
• Build and optimize inference pipelines to deliver real-time performance at scale
• Develop retrieval-augmented generation (RAG) systems and the surrounding data infrastructure
• Create evaluation frameworks that consistently measure AI quality and reliability
• Ship production APIs and the monitoring tooling needed to track model behavior in production
• Translate research ideas into shipped product features while balancing rigor and velocity

Skills

Software engineering

ML expertise

Transformer architectures

Production deployment

Inference optimization

Deployment patterns

Observability

Trade-offs

Job description

LockedIn AI is seeking a full-stack AI engineer to design, train, and deploy language and speech models for a real-time AI assistant. You will build end-to-end ML pipelines, from data prep to production monitoring, bridging research and production code.

The role requires hands-on ML with transformer models, optimization, and observability, ensuring rapid, robust shipping of features that impact users directly.