Nicholas Ali

Overview

Designed and implemented a multi-threaded HTTP web server capable of handling concurrent client connections. This project focuses on understanding the low-level mechanics of network protocols and system calls.

Problem

Standard web servers abstract away the underlying complexities of connection handling, request parsing, and memory management. Building one from scratch demonstrates mastery of these core systems concepts.

Architecture

Key Decisions

• Multi-threaded Architecture

  Utilized a thread pool to handle multiple connections simultaneously, preventing blocking on I/O operations.

• Custom HTTP Parser

  Wrote a zero-dependency HTTP parser to manually process request headers and bodies, ensuring strict compliance with HTTP/1.1 specs.

Tradeoffs

Performance

Code Example

1void Server::handleClient(int clientSocket) {
2    char buffer[4096] = {0};
3    read(clientSocket, buffer, 4096);
4    
5    Request req = parseRequest(buffer);
6    Response res = router.route(req);
7    
8    std::string responseStr = res.toString();
9    send(clientSocket, responseStr.c_str(), responseStr.length(), 0);
10    close(clientSocket);
11}

Overview

Built a sophisticated multi-agent system utilizing a supervisor architecture. The system dynamically selects the most appropriate LLM (phi3, gemma3, qwen3) based on task complexity and maintains long-term memory.

Problem

Single-agent systems often struggle with complex, multi-step tasks. This architecture decomposes problems and assigns them to specialized agents (Planner, Researcher, Executor) for higher accuracy.

Architecture

Key Decisions

• Supervisor Design Pattern

  Implemented a central orchestrator that delegates tasks to workers and aggregates results, ensuring coherent outputs.

• Dynamic Model Router

  Developed a routing logic that assigns lightweight models to simple tasks and powerful models to complex reasoning chains.

Tradeoffs

Performance

Overview

Developed the core systems of a game engine, including a rendering pipeline, physics simulation integration, and entity-component system (ECS). This project showcases low-level system design and performance optimization.

Problem

Commercial engines are powerful but obscure understanding of core game loop mechanics. Building an engine requires solving problems in memory management, linear algebra, and real-time rendering.

Architecture

Key Decisions

• Entity-Component System (ECS)

  Adopted an ECS architecture for data-oriented design, improving cache locality and performance for game objects.

• Physics Integration

  Implemented rigid body dynamics and collision detection, similar to industry standards found in Unreal Engine's Chaos.

Tradeoffs

Performance

Code Example

1void Game::run() {
2    while (isRunning) {
3        float deltaTime = calculateDeltaTime();
4        
5        physicsSystem.update(deltaTime);
6        inputSystem.processEvents();
7        
8        renderer.beginFrame();
9        renderSystem.draw(entities);
10        renderer.endFrame();
11    }
12}