Natural-Language Query Builder for Banking Data Templates

Added Nov 2025 3 design docs

Analysts and operations staff at banks constantly need data pulls, such as all dormant accounts opened before a date or monthly card spend by segment, but many cannot write the queries themselves and wait days for a technical colleague. A tool that turns a plain-English request into a vetted query template closes that gap. The intern builds this assistant as a streamlit application in python. A library of banking data query templates, each annotated with descriptions of what it retrieves, is embedded in a vector database. When a user types a request in natural language, a retrieval-augmented generation pipeline matches it against the template library, and generative AI adapts the closest template to the user's intent, explains what the query will return, and suggests clarifications when the request is ambiguous. Users can iterate conversationally, refining the request until the generated query matches what they meant, with RAG keeping every suggestion anchored to real, tested templates. The project teaches semantic matching between informal language and structured artifacts, prompt design for faithful adaptation, and interface patterns for human-in-the-loop refinement, which are practical skills for building AI copilots in data-heavy industries.

Related projects

Interactive Knowledge Graph Explorer for Banking Documents

Banks accumulate enormous libraries of reports, contracts, and statements, and the connections buried across those documents, such as which policies reference which entities, or how terms in one contract relate to clauses in another, are nearly impossible to see by reading files one at a time. Analysts and compliance staff need a way to explore that knowledge as a connected map rather than a pile of PDFs. The intern builds an exploration suite in python with a streamlit front end where users upload diverse banking documents. Text is chunked and embedded into a vector database, and a retrieval-augmented generation pipeline finds related passages across the corpus while generative AI produces explanations of how concepts, entities, and policies are linked. The centerpiece is an interactive map of key terms and their relationships that users can query in natural language, asking, for example, how a policy connects to a counterparty mentioned in a different report, with RAG grounding every answer in the source documents. The project is deliberately modular so a team can divide ingestion, retrieval, and interface work, and it leaves the intern with a portfolio-quality demonstration of embeddings, semantic retrieval, and grounded generation applied to a hard, realistic problem.

bankingbusinessgenerative-aipython+4

View project

Easy

Searchable Case Study Library for Banking Education

Case studies are one of the best ways to learn banking, since real events such as failed risk controls, successful product launches, and landmark regulatory actions teach lessons that textbooks flatten. The difficulty is discovery: relevant cases are scattered across sources, and a student rarely knows which one matches the topic at hand. The intern builds a streamlit application that makes a case study collection searchable by meaning rather than keyword. Case documents are embedded and stored in a vector database, and when a user asks about a topic such as credit risk in small business lending, a retrieval-augmented generation pipeline pulls the most relevant cases and generative AI produces a concise summary of each one, highlighting the key lessons and outcomes. Everything is written in python, and the interface lets users browse results, read summaries, and drill into the underlying case text. The project is an approachable introduction to RAG: the intern learns how embeddings capture topical similarity, how retrieval grounds generated summaries in real sources, and how to package the whole system as a clean educational tool for banking learners.

bankingresearchgenerative-aipython+4

View project

Easy

Beginner-Friendly Banking Concept Explainer with RAG Search

Terms like amortization, repo rate, and tier-one capital form a wall of jargon between newcomers and a working understanding of finance. Generic chatbots can define these terms but often hallucinate details, so a trustworthy explainer needs to ground its answers in a curated body of reference material. The intern builds that explainer as a streamlit application in python. A reference collection of banking definitions and explanatory articles is embedded into a vector database, and when a user types a concept they find confusing, a retrieval-augmented generation pipeline fetches the most relevant passages and generative AI composes a clear, beginner-friendly explanation from them, often with a simple example. Because every answer is grounded through RAG in the curated content, the tool avoids the invented facts that plague ungrounded chatbots, and users can see which sources informed each explanation. The project teaches the core mechanics of retrieval-augmented systems, including chunking, embedding, similarity search, and grounded prompting, while also exercising judgment about explanation quality. The intern finishes with a polished educational tool and a solid grasp of why grounding matters in AI products.

bankinggenerative-aipython+4

View project