The Fowler School of Engineering currently lacks the manpower to fulfill the advising needs of its students. Advisors spend a significant percentage of advising time handling common, basic questions which could be easily answered from existing documentation. In order to tackle this problem, we propose a locally hosted large language model leveraging a modern retrieval augmented generation pipeline in order to guarantee the language model generates the correct advising information for each student. Our RAG pipeline uses Qwen3.6:27b as the main LLM to generate responses, and a fine-tuned variant of Qwen3.5:4b as an intermediate LLM for miscellaneous tasks such as summarizing documents, query routing to collections, and message compression to preserve a manageable context window size. Two main databases are used, with a 4096 dimension Qdrant vector database used for retrieval and a PostgreSQL database used for memory. In addition, within the RAG pipeline, we have incorporated metadata filtering with the student’s catalog year, major, and minor in order to guarantee deterministic retrieval, hybrid search via both semantic dense search and sparse BM25 retrieval, and re-ranking via the BGE-Reranker-V2-M3 cross-encoder to improve retrieval results.
","abstract_html":"<p>The Fowler School of Engineering currently lacks the manpower to fulfill the advising needs of its students. Advisors spend a significant percentage of advising time handling common, basic questions which could be easily answered from existing documentation. In order to tackle this problem, we propose a locally hosted large language model leveraging a modern retrieval augmented generation pipeline in order to guarantee the language model generates the correct advising information for each student. Our RAG pipeline uses Qwen3.6:27b as the main LLM to generate responses, and a fine-tuned variant of Qwen3.5:4b as an intermediate LLM for miscellaneous tasks such as summarizing documents, query routing to collections, and message compression to preserve a manageable context window size. Two main databases are used, with a 4096 dimension Qdrant vector database used for retrieval and a PostgreSQL database used for memory. In addition, within the RAG pipeline, we have incorporated metadata filtering with the student’s catalog year, major, and minor in order to guarantee deterministic retrieval, hybrid search via both semantic dense search and sparse BM25 retrieval, and re-ranking via the BGE-Reranker-V2-M3 cross-encoder to improve retrieval results.</p>","abstract_has_math":false,"creators":["Au, Spencer"],"institution":null,"degree_name":"Master of Science (MS)","degree_level":"Thesis","degree_discipline":"Electrical Engineering and Computer Science","degree_department":null,"school":null,"contributors":["Jonathan D. Humphreys","Erik J. Linstead","Chelsea M. Parlett-Pelleriti","Elia E. Lledo"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-05-01T07:00:00Z","date_published":"2026-05-01T07:00:00Z","updated_at":"2026-07-24T01:38:47Z","subjects":["LLM","RAG","advising","undergraduate","human augmented ai","large language model","retrieval augmented generation","Artificial Intelligence and Robotics","Data Science","Systems Architecture"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://digitalcommons.chapman.edu/eecs_theses/11","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Jonathan D. Humphreys","Erik J. Linstead","Chelsea M. Parlett-Pelleriti","Elia E. Lledo"]},{"key":"dc:creator","label":"Author","values":["Au, Spencer"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2027-05-01T07:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical Engineering and Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science (MS)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["LLM","RAG","advising","undergraduate","human augmented ai","large language model","retrieval augmented generation","Artificial Intelligence and Robotics","Data Science","Systems Architecture"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.chapman.edu/eecs_theses/11"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The Fowler School of Engineering currently lacks the manpower to fulfill the advising needs of its students. Advisors spend a significant percentage of advising time handling common, basic questions which could be easily answered from existing documentation. In order to tackle this problem, we propose a locally hosted large language model leveraging a modern retrieval augmented generation pipeline in order to guarantee the language model generates the correct advising information for each student. Our RAG pipeline uses Qwen3.6:27b as the main LLM to generate responses, and a fine-tuned variant of Qwen3.5:4b as an intermediate LLM for miscellaneous tasks such as summarizing documents, query routing to collections, and message compression to preserve a manageable context window size. Two main databases are used, with a 4096 dimension Qdrant vector database used for retrieval and a PostgreSQL database used for memory. In addition, within the RAG pipeline, we have incorporated metadata filtering with the student’s catalog year, major, and minor in order to guarantee deterministic retrieval, hybrid search via both semantic dense search and sparse BM25 retrieval, and re-ranking via the BGE-Reranker-V2-M3 cross-encoder to improve retrieval results.
"]},{"key":"dc:source","label":"Dc Source","values":["S. Au, \"A large language model and retrieval-augmented generation system to assist with undergraduate academic advising,\" M. S. thesis, Chapman University, Orange, CA, 2026. https://doi.org/10.36837/chapman.000736"]},{"key":"dc:title","label":"Title","values":["A Large Language Model and Retrieval-Augmented Generation System to Assist with Undergraduate Academic Advising"]}]}],"canonical_facts":{"dc:contributor":["Jonathan D. Humphreys","Erik J. Linstead","Chelsea M. Parlett-Pelleriti","Elia E. Lledo"],"dc:creator":["Au, Spencer"],"dc:date.available":["2027-05-01T07:00:00Z"],"dc:description.abstract":["The Fowler School of Engineering currently lacks the manpower to fulfill the advising needs of its students. Advisors spend a significant percentage of advising time handling common, basic questions which could be easily answered from existing documentation. In order to tackle this problem, we propose a locally hosted large language model leveraging a modern retrieval augmented generation pipeline in order to guarantee the language model generates the correct advising information for each student. Our RAG pipeline uses Qwen3.6:27b as the main LLM to generate responses, and a fine-tuned variant of Qwen3.5:4b as an intermediate LLM for miscellaneous tasks such as summarizing documents, query routing to collections, and message compression to preserve a manageable context window size. Two main databases are used, with a 4096 dimension Qdrant vector database used for retrieval and a PostgreSQL database used for memory. In addition, within the RAG pipeline, we have incorporated metadata filtering with the student’s catalog year, major, and minor in order to guarantee deterministic retrieval, hybrid search via both semantic dense search and sparse BM25 retrieval, and re-ranking via the BGE-Reranker-V2-M3 cross-encoder to improve retrieval results.
"],"dc:identifier":["https://digitalcommons.chapman.edu/eecs_theses/11"],"dc:source":["S. Au, \"A large language model and retrieval-augmented generation system to assist with undergraduate academic advising,\" M. S. thesis, Chapman University, Orange, CA, 2026. https://doi.org/10.36837/chapman.000736"],"dc:subject":["LLM","RAG","advising","undergraduate","human augmented ai","large language model","retrieval augmented generation","Artificial Intelligence and Robotics","Data Science","Systems Architecture"],"dc:title":["A Large Language Model and Retrieval-Augmented Generation System to Assist with Undergraduate Academic Advising"],"thesis:degree_discipline":["Electrical Engineering and Computer Science"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["Master of Science (MS)"]},"updated_at":"2026-07-24T01:38:47Z"}