Nexlevr
AI Fluency & Building · Internship Role

AI KNOWLEDGE SYSTEMS

Build a searchable threat-intel knowledge base. Collect, clean, chunk and retrieve information so AI answers with real sources.

SKILLS YOU'LL WALK AWAY WITH
RAGEmbeddings & vector searchData cleaningMetadata designThreat-intel formats (STIX)
PROOF-OF-WORK TASKS
8 TASKS

Start at Beginner and work up. Every task produces something real you can show: a report, a tool, a guide or a demo. Publish it on LinkedIn or GitHub as your proof of work.

B
Beginner
LEVEL 01
  • 01Collect 30 public cybersecurity advisories or articles. Organise them in a spreadsheet with title, date, source, type and tags, and explain your tagging scheme.
  • 02Explain embeddings to a non-technical friend with a one-page visual and a tiny demo showing similar sentences sitting close together.
  • 03Build a 'keyword vs meaning' demo: search 50 documents with keyword search and with embedding search, and show 3 queries where the results differ.
I
Intermediate
LEVEL 02
  • 01Build a RAG chatbot over MITRE ATT&CK techniques that answers questions and cites the technique ID for every claim.
  • 02Test 3 chunking strategies (fixed size, by heading, by paragraph) on the same documents. Measure which one retrieves the right passage most often across 20 questions.
  • 03Design a metadata schema for a threat-intel knowledge base (source, date, confidence, TLP, actor, sector) and load 100 records into a vector database.
A
Advanced
LEVEL 03
  • 01Build a full threat-intel search system with daily ingestion of CISA KEV and news feeds, de-duplication, hybrid search, cited answers, and an honest 'I don't know' when sources are missing.
  • 02Create a 30-question evaluation suite with gold answers for your knowledge base. Measure retrieval and answer accuracy, and publish a report on what improved them.