Problem
Chess positions create a high-dimensional comparison problem where exact matches are less useful than strategically similar examples.
LiveMachine learningBackend engineeringData systems
SIMBA
SIMBA explores how representation, indexing, and ranking quality come together in a search-heavy technical system. It is useful portfolio signal because it connects algorithms, data representation, backend thinking, evaluation, and applied AI into one explainable project.
Signal
Large search space
Signal
Ranking evaluation
Signal
Interactive querying
Case study
What this project demonstrates
Approach
- Represent board states with features that preserve useful decision context.
- Index the search space for fast nearest-neighbor retrieval.
- Evaluate retrieved positions with ranking metrics and qualitative chess relevance.
Outcome
- Shows algorithmic thinking across representation, retrieval, ranking, and service design.
- Creates a foundation for a public demo where users can query positions and inspect similar results.
- Now has an always-on portfolio landing page so the project remains accessible even when the free Streamlit demo sleeps.
Stack
Technology used or planned
Tooling belongs here, where it supports the project story instead of crowding the homepage hero.
PythonFAISSPyTorchAWS