Open to opportunities

Muhammad faisal

@muhammadfaisal20

Message

I benchmark and refine LLMs while building reliable backend evaluation systems.

Pakistan

Message

What I'm looking for

I'm looking to build and evaluate intelligent software systems, especially LLM-driven products, backend services, automated testing pipelines, and scalable deployment environments.

At Next AI, I train and refine next-generation LLMs using human conversations, improving GPT-generated web application output quality by 45% through iterative alignment feedback.

I evaluate AI-generated applications end to end, from multi-turn webpage generation to SWE-Bench code fixes in Python and Java. I’ve manually refactored generated frontend and backend code to reach 98% prompt compliance and functionality, and identified 50+ critical failure modes across Anthropic, Gemini, and OpenAI models.

I also build automated unit test suites, backend microservices, and Dockerized evaluation environments for reproducible AI code-fix pipelines.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Next AI

Aug 2024 - Present (2 years)

Trained next-generation LLMs (GPT-5) using human conversations, improving output quality by 45% through iterative alignment feedback. Conducted multi-turn evaluations and benchmarked LLM performance on SWE-bench, enhancing autonomous coding accuracy by 40%.