AI Engineer Intern: Production ML, Benchmarks & Infra

Thommidhi; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

Thommidhi is seeking an engineer to deploy LLMs and integrate Claude with open-source models via a unified interface. You will run benchmarks and optimize simulation systems on Cloudflare Workers, Workflows, and D1 to improve speed and reliability.

You will implement retry logic, output validation, and data lineage tracking while building safeguards against nonsensical outputs and establishing robust monitoring to compare predictions with real-world results.

Qualifications

• Experience deploying an LLM in production.
• Proficiency in TypeScript or Python for production code.
• Ability to critically evaluate model outputs rather than accepting them at face value.

Responsibilities

• Integrate Claude and open-source models via a unified interface for interchangeable use.
• Conduct comparative benchmarks across scenarios to measure model performance.
• Improve simulation systems built on Cloudflare Workers, Workflows, and D1 for speed and reliability.
• Add retry logic, output validation, data lineage tracking, and spending controls.
• Develop safeguards to prevent nonsensical numerical outputs.
• Set up monitoring and logging to track predictions against real-world outcomes.

Skills

LLM deployment

TypeScript

Python

Model output verification

AI infrastructure

Monitoring & logging

Cost controls

Education

Bachelor's degree in CS/engineering or related field

Tools

Cloudflare Workers

Cloudflare D1

Cloudflare Workflows

Job description

Thommidhi is seeking an engineer to deploy LLMs and integrate Claude with open-source models via a unified interface. You will run benchmarks and optimize simulation systems on Cloudflare Workers, Workflows, and D1 to improve speed and reliability.

You will implement retry logic, output validation, and data lineage tracking while building safeguards against nonsensical outputs and establishing robust monitoring to compare predictions with real-world results.