Willingness to travel to various locations to support the SEI’s overall mission. This includes within the SEI and CMU community.
You will be subject to a background check and must be eligible to work in the United States without Visa sponsorship.
Duties
Assist in designing and prototyping autonomous agents and multi-step pipelines using frameworks such as LangChain, ReAct, tool-calling, or the Model Context Protocol (MCP).
Build and maintain data pipelines, ETL processes, and backend services using Python, C/C++, or Java.
Contribute to Retrieval-Augmented Generation (RAG) pipelines that integrate external knowledge bases with Large Language Models (LLMs) to improve factual accuracy.
Support rapid prototyping, benchmarking, and robustness or adversarial testing of AI models in realistic environments.
Research and stay current with emerging LLM architectures, agentic paradigms, parameter-efficient fine-tuning (PEFT/LoRA) methods, and AI-safety techniques.
Knowledge, Skills and Abilities
Proficiency in Python and familiarity with at least one compiled language (such as C/C++ or Java), REST/GraphQL APIs, or containerization.
Foundational understanding of machine learning theory (supervised, unsupervised, and reinforcement learning) and standard evaluation metrics.
Academic or practical experience working with deep learning frameworks (such as PyTorch, TensorFlow, or JAX) and LLM libraries (such as Hugging Face Transformers or LangChain).
Analytical thinking with the ability to decompose complex AI problems into tractable components and iterate rapidly.
Strong written and verbal communication skills for documenting designs and presenting results to technical and non-technical stakeholders.
Proven teamwork, high curiosity, and autonomy to collaborate in interdisciplinary groups, contribute to shared codebases, and explore emerging technologies.
Desired Experience
Experience fine-tuning LLMs or building multi-tool agents, planner-executor loops, or tool-calling pipelines through coursework, research, or personal projects.
Exposure to deploying ML/LLM APIs (e.g., FastAPI, gRPC) or working with data-pipeline tools (e.g., Airflow, Prefect, Ray).
Familiarity with building evaluation frameworks, benchmarks, or data quality pipelines for machine learning or agentic systems.
Contributions to open-source AI/ML libraries, agentic frameworks, or prior research experience relevant to defense or government AI applications.
Job Function Breakdown
50% AI Prototyping & Software Engineering: Assisting in the design, coding, and deployment of agentic workflows, RAG pipelines, and backend services in Python, C/C++, or Java.
30% Applied Research, Benchmarking & Testing: Conducting literature reviews on emerging LLM/agentic paradigms, running benchmarks, and assisting with robustness and adversarial evaluations.
10% Documentation & Knowledge Sharing: Documenting experimental designs, codebases, and evaluation results, and presenting findings to team members and stakeholders.
10% Team Collaboration & Mentorship: Collaborating with senior researchers and engineers to refine problem statements, participating in code reviews, and engaging in SEI learning opportunities.
TOTAL = 100%
Location
- Pittsburgh, PA
- Job Function
- Non-CMU Students
- Position Type
- Intern (Fixed Term)
- Full time/Part time
- Full time
- Pay Basis
- Hourly
- More Information:
Please visit “Why Carnegie Mellon” to learn more about becoming part of an institution inspiring innovations that change the world.
Click here to view a listing of employee benefits
Carnegie Mellon University is an Equal Opportunity Employer/Disability/Veteran.
Statement of Assurance