Projects
Projects
Production ML systems with measured results. Each project includes a full write-up on this site.
RelevanceIQ: Grounded Product Search
Hybrid retrieval, cross-encoder reranking, and citation-grounded RAG over 568K Amazon reviews.
| Stack | OpenSearch, FAISS, FastAPI, MLflow, Ollama |
| Impact | +17.7% NDCG@10, 97.5% faithfulness, sub-80ms p95 search |
| Write-up | Read the post |
KAIROS: Decision Intelligence System
Production decision engine with hybrid ensembles, isotonic calibration, and full observability.
| Stack | LightGBM, CatBoost, FastAPI, Prometheus, Grafana, Docker |
| Impact | 96% precision, 70% automation, ECE < 0.02 |
| Write-up | Read the post |
Brewing Intelligence
Structured DSA notes with pattern-based solutions, four curated interview sheets (Blind 75, Grind 75, LeetCode Top 150, NeetCode 150), and Python reference material.
| Stack | MkDocs Material, Python |
| Content | 23 DSA pattern pages, 4 interview sheets (543 problems), Python tutorials |
| Live site | abhijitmore.github.io/BrewingIntelligence |
DSA Venn Explorer
Interactive explorer for 300 consolidated interview questions across Blind 75, Grind 75, LeetCode Top 150, and NeetCode 150 — with Venn overlap filtering and links to Brewing Intelligence solutions.
| Stack | Vanilla JS, Python data pipeline |
| Features | 4-set Venn diagram, category/tag filters, solved tracking, solution links |
| Live app | abhijitmore.github.io/DSA-Venn-Explorer |
More on GitHub
Other repos including early RAG experiments, edge inference work, and research notes are on github.com/AbhijitMore.