AI Internship: Build Local, Self-Hosted LLM Pipelines

ELCA Switzerland and Europe; Zürich; On-site; CHF 20,000 - 28,000; Full time

Benefits offered by this job

Monthly AI meetups

Hackathons and Brownbag sessions

Technical blog

Monthly After-Works

Job summary

ELCA in Switzerland offers a 6-month internship starting Spring 2027 to explore sovereign, locally-run AI for complex legacy data migrations. You will benchmark open-source LLMs, deploy models locally with Ollama, vLLM, or Hugging Face, and design a prototype pipeline that analyzes legacy schemas, code and sample data.

You will uncover hidden business rules in legacy systems (including .NET apps), validate data quality, and generate automated effort estimates.

Qualifications

• Bachelor or Master degree in Computer Science or related discipline.
• Proficient in Python and SQL; familiarity with .NET-based systems is a plus.
• Experience with LLMs and local deployments (Ollama, vLLM, Hugging Face).
• Interested in agentic AI and building benchmark pipelines.
• Basic understanding of database schemas and data quality.
• Structured work approach and clear communication.
• Fluent in written and spoken English, plus German or French.

Responsibilities

• Benchmark open-source LLMs for self-hosted vs cloud models on migration tasks.
• Deploy candidate models locally using Ollama, vLLM, or Hugging Face.
• Design and build a prototype pipeline that analyzes legacy schemas, code and data.
• Uncover hidden business rules in legacy systems, including .NET apps.
• Validate data consistency and assess complexity of a migration case.
• Generate automated effort estimates and migration guidelines from a local model.
• Define criteria for when a local model can replace manual scoping in bids.
• Evaluate trade-offs between self-hosted and cloud models for licensing and cost.
• Shape how we scope fixed-price services for data sovereignty needs.

Skills

Python programming

SQL

.NET familiarity

LLMs

model deployment

data schemas

data quality

agentic AI

German or French

Education

Bachelor or Master in Computer Science

Tools

Ollama

vLLM

Hugging Face

Job description

ELCA in Switzerland offers a 6-month internship starting Spring 2027 to explore sovereign, locally-run AI for complex legacy data migrations. You will benchmark open-source LLMs, deploy models locally with Ollama, vLLM, or Hugging Face, and design a prototype pipeline that analyzes legacy schemas, code and sample data.

You will uncover hidden business rules in legacy systems (including .NET apps), validate data quality, and generate automated effort estimates.