Optifiner is a self-evolving code framework that automatically improves codebases through multi-agent AI-driven optimization. It spawns parallel AI agents that propose and test code improvements, keeping only changes that measurably improve performance against benchmark metrics.

A modular quantitative backtesting framework built from scratch in Python — featuring data pipelines, signal generation, execution simulation, performance analytics, and unit tests. Includes notes on extending the system for ML/AI-driven trading strategies.

AI-driven performance testing framework that automates the complete journey from real user interaction to protocol-level load testing. The framework integrates Claude AI, Playwright, Apache JMeter, and the Model Context Protocol (MCP) to eliminate manual scripting and enable intelligent orchestration, execution, and analysis.

Agnox | High-performance, framework-agnostic test automation platform. Run any containerized suite (Playwright, Pytest, etc.) with real-time logs, multi-tenancy, and AI-driven failure analysis.

Designed a rules-driven QA automation framework using a single OpenAPI spec as the source of truth. Automated API, UI, and integration tests with AI-generated BDD scenarios. Built an MCP server to control Playwright, validated API-UI data consistency, added rule-based reporting, k6 performance testing, and CI/CD-ready scalability.

Minimal Capability Design (MCD) framework for edge AI agents. MCD uses stateless, prompt-driven architectures with quantized models (Q1/Q4/Q8) rather than memory-heavy systems. Simulations and tests validate constraint-first design achieves sufficient performance for IoT deployment where traditional agents fail.