Lewis Coding LogoLewis CodingAndroid and Web development
HomeWebPortfolioApps

Lewis Coding

Web development, Android apps, and practical software for growing businesses.

Privacy Policies
© 2026 Joshua Lewis
Back to project catalog

Internship project

Infoseek

A full-stack enterprise RAG prototype built during my internship at Infosys, designed to replace traditional intranet search systems.

Next.jsPythonGCPAlloyDBBig Query
Private Codebase
Blue Infoseek wordmark on a white background

About the project

Built independently during my internship at Infosys, Infoseek explores a future where employees can ask questions using natural language instead of manually searching across fragmented knowledgebases.

It ingests PDFs and websites, preserves document structure while chunking content, creates semantic embeddings, and retrieves the most relevant evidence to ground each AI-generated answer.

The prototype pairs a polished Next.js interface with a Python and Flask retrieval pipeline, local Qwen inference, role-aware access controls, and GCP-backed vector search—anambitious foundation for an enterprise knowledge layer that could reduce dependence on conventional web search.

Highlights

  • Prototype built from the ground up independantly.
  • Implements both Naive and HyDE RAG pipelines
  • Built with big data principles using a range of the most effective GCP services.

Project gallery

Infoseek home page with the Infoseek logo above a central search bar
Infoseek search landing page
Infoseek insights dashboard with usage metrics and a user queries line chart
Search and engagement insights dashboard
Infoseek insights dashboard showing a 3D semantic topic visualization alongside topic details and queried documents
Semantic topic clustering and query insights
Infoseek search results page showing an AI-generated answer, source chunks, feedback controls, and source documents
AI answer with retrieved source evidence
Infoseek upload documents page showing indexed files and a successful upload confirmation
Document ingestion and indexing dashboard