Amber Taveras
Software Engineer
Specializing in C++ and Rust
#include <about.h>
A systems-focused software engineer with experience in low-level programming, infrastructure, and numerical computing in C++ and Rust. Pursuing expertise in computer graphics rendering, memory management, and multithreaded systems. Based in the New York Metropolitan Area.
#include <skills.h>
- C++
- Rust
- Python
- TypeScript
- Docker
- AWS
- PostgreSQL
- MongoDB
#include <experience.h>
-
Software Intern
MTA, Metro-North Railroad
October 2025 - June 2026
- Developed a real-time dispatching simulator in Rust to simulate thousands of concurrent trains, using physics-based movement and Tokio async runtimes on multithreaded processes decoupled via Unix sockets.
- Configured a virtualized test infrastructure using Proxmox and Linux KVM, provisioning 8 Windows VMs joined to an Active Directory domain controller to automate Windows Event Log and Event Trace forwarding via Windows configurations and custom Powershell scripts.
- Built data aggregation tools in Rust and Python to calculate cumulative wear and tear on field infrastructure from operational logs and SQL databases, producing Excel reports and graph visualizations.
- Developed a high-throughput log aggregation tool in Go using goroutines and concurrent processing to ingest and structure gigabytes of daily log data from multiple virtual machines into a VictoriaLogs time-series database.
- Rust
- Golang
- Python
- Linux
- Linux KVM
- Proxmox
- Windows Server
- Powershell
-
Software Engineering Intern
StudyFetch
January 2025 - May 2025
- Developed a generative 2D/3D diagram engine in TypeScript and Python with interactive features to aid learning.
- Built a 3D racing game in Three.js featuring procedurally generated racetracks, NPCs, and real-time physics-based collision detection.
- Designed a parser to transform structured responses from OpenAI prompts into concept map nodes for study aids.
- TypeScript
- Next.js
- Python
- Three.js
#include <projects.h>
-
Memory Allocators
A custom memory allocator library in modern C++, implementing linear, pool free list, and buddy allocation strategies. Each allocator is benchmarked against one another and the standard implementation to assess speed, fragmentation, and flexibility. Includes full API documentation and a WebAssembly (WASM) visual component compiled with Emscripten to visualize the C++ allocator state in the browser.
- C++
- CMake
- Google Test
- Google Benchmark
- Emscripten
- WebAssembly (WASM)
-
3D Software Renderer
A CPU-based 3D software renderer in C++, with no GPU graphics dependencies. The renderer implements a custom rasterizer to convert 3D geometry into 2D pixel output, along with projection and view matrix calculations to support a fully orbiting camera and 3D-to-2D transforms. The rasterizer outputs a framebuffer to pass into SDL2, which handles only windowing and pixel output.
- C++
- SDL2
- CMake
-
Asset Portfolio Optimizer
An asset management portfolio optimizer that implements the cross-validated exponentially weighted average sample covariance estimator (EWA-CV) detailed in Estimation of Large Financial Covariances: A Cross-Validation Approach to construct minimum variance portfolios from real market data. Developed in collaboration with Baruch College's Financial Quants and Engineers Club.
- C++
- CMake
- Pybind11
- Python
- Yahoo Finance
#include <contact.h>
Connect with me