Adi, Azarel Grahandito (2025) Implementasi Sistem Agen AI Interaktif Berbasis LangChain dengan Grok dan GPT untuk Analisis dan Visualisasi Data pada Basis Data Relasional. Project Report. [s.n.], [s.l.]. (Unpublished)
|
Text
5025221126-Project_Report.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Kerja Praktik ini mengembangkan sebuah Proof of Concept (PoC) untuk PT. Astra Graphia Information Technology (AGIT), berupa sistem Agentic AI yang mampu mengubah pertanyaan bahasa alami menjadi insight berbasis data. Dengan memanfaatkan Microsoft Azure AI Foundry untuk inference dan LangChain untuk orkestrasi AI, sistem ini mengotomatiskan analisis melalui generasi SQL, eksekusi kueri, dan visualisasi data. Arsitektur ini dibangun di atas backend FastAPI dan antarmuka Streamlit untuk menjawab kebutuhan industri akan solusi analitik AI yang canggih dan mudah diakses. Fitur utama sistem ini meliputi perencanaan multi-kueri untuk pemahaman konteks data yang mendalam dari database dan kemampuan memilih jenis visualisasi secara mandiri. Alur kerja agen mencakup analisis maksud, pembuatan SQL yang tervalidasi skema, eksekusi pada berbagai sumber data, dan peringkasan hasil. Kontribusi utamanya adalah desain arsitektur yang dapat diperluas dan pipeline end-to-end yang mempercepat eksplorasi data tanpa perlu mengekspos keseluruhan database perusahaan, sehingga menjadi portofolio solusi yang robust dan aman.
====================================================================================================================================
This internship developed a Proof of Concept (PoC) for PT. Astra Graphia Information Technology (AGIT) in the form of an agentic AI system capable of transforming natural language questions into data-driven insights. By leveraging Microsoft Azure AI Foundry for inference and LangChain for AI orchestration, the system automates analysis through SQL generation, query execution, and data visualization. The architecture is built on a FastAPI backend and a Streamlit interface to meet industrial needs for advanced and easily accessible AI analytics solutions. The main features of the system include multi-query planning for deeper contextual understanding of database information and the ability to autonomously select visualization types. The agent workflow covers intent analysis, schema-validated SQL generation, execution across multiple data sources, and summarization of results. The main contribution lies in the extensible architecture and end-to-end pipeline that accelerates data exploration without exposing the company’s entire database, making it a robust and secure solution portfolio.
| Item Type: | Monograph (Project Report) |
|---|---|
| Uncontrolled Keywords: | Agentic AI, LangChain, SQL, Microsoft AI Foundry, LLM, Analisis Data |
| Subjects: | Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Azarel Grahandito Adi |
| Date Deposited: | 08 Dec 2025 07:52 |
| Last Modified: | 08 Dec 2025 07:52 |
| URI: | http://repository.its.ac.id/id/eprint/128880 |
Actions (login required)
![]() |
View Item |
