An End-to-End Evaluation Framework for Entity Resolution Systems

Save up to 40% on agent token costs with code graphs: call graphs, dependency graphs, dead code detection, and blast radius analysis.

A single-file Python CLI that pre-registers AI/ML accuracy claims with SHA-256. Lock the threshold before the data, or it didn't happen.

A metrics library to evaluate vision language models with a pytorch eco system.

✍️ Collaborate on writing technical content for the Giskard Community

An open-source Streamlit web app to generate beautiful confusion matrices for multi-class machine learning models. Supports numeric and string labels, CSV upload, manual label entry, custom color maps, and displays evaluation metrics like Accuracy, Precision, Recall, and F1-score. Users can download the confusion matrix as an image.

Safety-first legal NLP system with hierarchical long-document processing, deterministic inference, clause extraction, and rule-based risk engine — built for traceability and deployment constraints.

Measurement-disciplined optimizer for Claude Agent Skills — three-gate system (stability, effect size, function preservation), anchored rubric, train/holdout split. Inspired by Karpathy's autoresearch and alchaincyf's darwin-skill.

Enterprise-grade machine learning observability platform that detects data drift, concept drift, and performance degradation in production models. Features statistical drift detection (KS test, PSI), real-time alerting, Redis caching, and FastAPI backend.

Data Science Challenge from Coursera Project : Loan Default Prediction

This project contains codes and paperwork based on the course CSI5155 at University of Ottawa (delivered by Professor Dr. Herna Viktor).

Local-first stratified-sample audit UI for ML classifiers and labeled datasets. Wilson CIs, keyboard-first, no cloud.

GitHub Action for SWE-bench Pro evaluation powered by mcpbr

Small, educational project that shows how to build a **minimal RAG pipeline** with a **simple evaluation loop**

Public scorecard of how 25+ ML eval claims meet 9 PRML falsifiability criteria. CC0 data; MIT tooling.

End-to-end E-Commerce Recommendation System using implicit feedback, featuring Popularity, Item-Item CF, ALS (Matrix Factorization), and a Hybrid model, with offline evaluation and online serving via FastAPI + Streamlit.

Evaluation of system-level risks in content moderation models using policy-driven metrics, identity-based analysis, and governance-aligned datasets.

Static HTML backtest reports from eval JSON (calibration, Brier, CLV, optional bet ledger)

Collection of Machine Learning (ML) and Natural Language Processing (NLP) projects showcasing a range of applications, algorithms, and techniques.

A decision-oriented benchmark framework for evaluating action-conditioned world models beyond static AI benchmarks.