I build reliable AI products end to end.

Profile picture

Joshua Kirby

  • Full‑stack TypeScript / Next.js with LLM‑backed products
  • Native in Japanese and English

I’m a 2nd‑year AI & CS student with experience shipping production-grade LLM applications for real-world businesses.

I like taking ambiguous ideas and turning them into reliable AI products.

Work Experience

  • GenkiJACS

    Sep 2023 – Jun 2024

    Software Engineer Intern

  • The HALO Trust

    Dec 2021 – Apr 2023

    Student Intern - Technology

Education

  • University of Birmingham

    • B.Sc. Artificial Intelligence & Computer Science
    • 2nd year (expected graduation 2027)

GenkiJACS Internship

  • RAG chatbot handling 10,000+ messages from 2,000+ users, saving ~70 staff‑hours/week.

  • Zero‑maintenance scraping pipeline that provides analytics to guide site improvements.

  • Conducted ablation studies with 700+ relevance judgments to optimise the RAG pipeline.

Read Case Study

AI Conversational Placement Test

  • Cut the placement process from 2 weeks to 30 minutes by replacing manual Zoom interviews.

  • Automated a senior teacher's grading process with a state machine validated by a 600-sample eval suite.

  • Designed a multi-tenant platform with configurable auth that grants ownership while supporting diverse privacy requirements.

Read Case Study

View site source