Berke Kara

Berke Kara

I build backend systems and AI tools.

I’m a computer science senior at UT Austin, expecting to graduate in May 2027. I build research software at Idaho National Laboratory. I also co-develop robotics simulation tools with a graduate researcher in Dr. Lili Qiu’s group.

I’m looking for new-grad software engineering roles in backend systems, applied AI, or infrastructure. I’m open to opportunities anywhere in the United States.

Experience

  1. 2026–present

    Idaho National Laboratory

    Software Engineer Intern, Project Lead

    I work with researchers to figure out what they need, lead development of the platform, and help keep it running in production.

    Dr. Lili Qiu’s group, UT Austin

    Undergraduate Researcher

    I’m co-developing a robotics simulation harness with a graduate researcher using Isaac Sim and Isaac Lab. We’re testing navigation, manipulation, and the interface between the robot and the agent.

    Texas Cybersecurity Clinic

    Cybersecurity Student Analyst, Project Lead

    I lead a three-student team working with a Texas school district. The district adopted the acceptable-use, AI, and software-approval policies we delivered.

  2. Summer 2025

    Idaho National Laboratory

    Software Engineer Intern, Project Lead

    I led development of an app for asking questions about laboratory documents. It brought vector and graph retrieval together behind one interface.

Selected projects

Nuclear Data Explorer

Research data tools

I lead development of a browser-based tool that helps 30 INL researchers compare nuclear data across 12 evaluated libraries.

I cut the time to finish three concurrent queries from 9.0 s to 3.0 s in a benchmark.

Open project
Details

The challenge was keeping the FastAPI app responsive while a Java tool handled the queries. I moved blocking work to threads, limited concurrent Java processes, and capped the cache’s memory use.

I test concurrent requests and caching, and compare the analytics with saved reference results.

LuminRAG

RAG & Applied AI

I’m the main author of LuminRAG, a study tool that brings lecture videos, audio, PDFs, and slides into one searchable knowledge base. It connects ideas in a concept graph and answers questions with links back to the source material.

Open project GitHub

Requires your own AI provider API key.

Details

The main challenge is finding useful context across different kinds of course material. LuminRAG can search for relevant passages, follow connections in the graph, or combine both.

Tests cover the ingestion and retrieval workflows, and citations let readers check the sources behind an answer. It’s still a research prototype, with an unpublished independent research paper.

Comparative RAG Databases

Retrieval systems

I led development of this document Q&A app during my 2025 internship at INL. It brings Qdrant, ChromaDB, and Neo4j into one interface so researchers can compare different retrieval approaches.

Open project
Details

The backend retrieves passages from the selected databases, reranks them with a cross-encoder, and streams an answer as it’s generated. The repository includes setup instructions for running the app and database containers locally.

My Kitchen

Full-stack application

My Kitchen brings recipes, pantry inventory, and shared grocery lists into one place. I built the SvelteKit frontend and PostgreSQL backend, including barcode scanning.

For a 40-item grocery-list refresh, I cut the SQL statements from 130 to 13 in local measurements.

Open project
Details

The tricky part is shared inventory: two people can update it at the same time, and a request might arrive twice. I use operation IDs and database transactions to avoid counting the same change twice.

Database tests check concurrent updates, and browser tests cover the app’s everyday workflows.

Contact

The best way to reach me is berkekara@utexas.edu. You can also reach me at karaberke001@gmail.com.

UT Austin, B.S. Computer Science. Expected May 2027.