AI Voice Generation Tool
Production-grade TTS pipeline with sub-200ms latency
Systems and products built to solve real problems at scale. Each one taught me something new about the craft.
Production-grade TTS pipeline with sub-200ms latency
Real-time visibility across 40+ services
Habit-building platform with streak analytics
A selection of open-source experiments and projects exploring AI, machine learning, and systems engineering. Full collection at chemacabeza/my-github-projects.
Train in Python → Export to ONNX → Serve in Java
A complete, production-ready machine learning pipeline that trains a neural network in Python (PyTorch), exports it to ONNX, and serves it via a Spring Boot API with 'Golden Parity' tests. This automated architecture ensures mathematically identical results between Python and Java runtimes, proving that models can be trained in Python but executed with zero-Python production footprints.
Hybrid Architecture · Online & Offline Video Synthesis
A full-stack video platform supporting 5 state-of-the-art models (Wan, Kling, LTX-2, PixVerse) via fal.ai. Engineered with a flexible hybrid architecture: a single configuration switch lets you toggle between running models entirely offline on local hardware or connecting to cloud APIs for scale. Features async job orchestration, isolated Docker environments, and zero-downtime database migrations.
Text-to-speech Generator · OpenAI TTS
Convert text to natural-sounding speech instantly using the OpenAI TTS API. Choose from 11 expressive voices and stream raw MP3 audio straight to your browser. The secure architecture keeps API keys strictly server-side, while the Spring Boot backend and React frontend are fully Dockerized for effortless, horizontal scaling in any environment.
Whisper STT · GPT-4o Chat · OpenAI TTS
A production-ready voice assistant featuring speech-to-text (Whisper), intelligent GPT-4o chat responses, and natural TTS output. Includes configurable wake-word detection ('Nova'), support for 9 languages, and custom AI personas. Built with a premium dark-themed React UI and a Spring Boot backend, completely containerized for instant local deployment via Docker.
Hybrid Architecture · Online & Offline Transcription
A high-performance transcription pipeline powered by OpenAI Whisper. This hybrid platform allows you to toggle instantly between local-only transcription for air-gapped privacy or cloud acceleration via OpenAI APIs. Featuring a Next.js 15 App Router frontend and a Spring Boot 3.3 backend with PostgreSQL, it includes searchable history, file uploads, and browser-based recording updates.
Conversational AI · Whisper + GPT-4o + TTS
"Hey Nova, tell me a joke." Speak naturally, and this full-stack AI assistant listens (Whisper), thinks (GPT-4o), and speaks back (TTS) — all in real time, right in your browser. Features wake word detection, 9 languages, 6 voice personas, custom AI naming, push-to-talk and continuous listening modes, and both online and offline backends. It's your own Alexa — but you built it, you control it, and it runs with a single Docker command.
A Developer's Guide · 7-part living book
A structured deep-dive into Bash systems engineering — from variables and control flow to process management and signals. This 7-part living book is written for engineers who want to go beyond quick scripts and truly master the shell that holds systems together. Features chapters on subshells, Coprocesses, and Programmable Completion.
24-part deep dive · 70+ projects · 133+ tests
A massive mastery curriculum split into two deep tracks. The 13-part Java track spans OOP essentials, the Streams API, Concurrency, and JNI/Project Panama. The 11-part Spring Boot track covers REST, Security, JPA internal mappings, and advanced architectural patterns including Spring Modulith, event-driven design, and the Model Context Protocol (MCP) for LLM tool integration. The curriculum features a catalog of 70+ hands-on projects and a completely Dockerized Documentation Suite, guaranteeing instant, reproducible local execution across Mac and Linux.
Stable Diffusion · Fooocus · Model Library
A local high-performance art platform using Stable Diffusion and Fooocus. Features a 46GB model library with 7 SDXL checkpoints and 41 LoRAs, optimized for Apple Silicon via MPS GPU acceleration. Includes a centralized storage architecture and pre-configured workflows for production-quality character and environment synthesis.
20 chapters · 7 phases · Theoretical & Distributed AI
A comprehensive, senior-level systems engineering guide diving into the deepest technical layers of modern artificial intelligence. Spans 7 phases focusing on how massive neural networks are compiled and accelerated via hardware, decentralized via Federated Learning, and secured at scale. It extends into Graph Neural Networks, modern Natural Language pipelines, Autonomous Vehicle Perception, Biometric Security, Symbolic AI (Logic Programming & PDDL), AI Agents (RAG), and a deep exploration of Deep Learning Foundations exploring both TensorFlow and PyTorch architectures from scratch.
21 chapters · Hands-on · Fully Dockerized exercises
A hands-on, beginner-friendly curriculum synthesized from over 50 academic textbooks spanning 21 chapters of modern AI. Every chapter includes a fully Dockerized exercise — no local setup beyond Docker. Covers neural networks from scratch, Federated Learning, Reinforcement Learning, NLP, SVMs, Swarm AI, Computer Vision (CNNs), Transfer Learning, Semi-Supervised Learning, Autonomous Vehicles, Algorithmic Trading, Graph Neural Networks, DQN for games, Active Learning, Adversarial ML defenses, Fake News Detection, IoT Edge AI, Lifelong Learning, Metric Learning, ML Data Pipelines, and Game Theory.
20 chapters · Red & Blue Team · Dockerized exercises
A comprehensive, hands-on curriculum synthesized from 24 leading industry textbooks. Takes you from fundamental defensive principles and system hardening to advanced offensive red team operations. Includes fully Dockerized exercises covering Cryptography, Penetration Testing methodologies, Web App Security, Malware Analysis, Reverse Engineering with Ghidra, Digital Forensics, Cloud Security, and Security Automation.
Advanced Systems Programming · Mastery Sequence
A 12-part sequence in high-performance C++ systems engineering. Covers the C++ Object Model under the hood, memory architecture, virtual tables, and template metaprogramming (SFINAE). Explores policy-based design and assembly-level reversing, culminating in a Mastery Project focusing on type-erased thread-safe containers.
Microservices & Concurrency · 12-part guide
A comprehensive path from zero to production Go. Mentions the CSP concurrency model (Goroutines/Channels), Test-Driven Development (TDD), and Domain-Driven Design (DDD). Culminates in building multi-container microservices communicating over gRPC with Protocol Buffers, fully Dockerized for containerized local execution.
34 chapters · 8 phases · Scale to Millions
How do you design YouTube? How does WhatsApp deliver a message in 30 ms? This 34-chapter curriculum, distilled from over 30 professional-grade textbooks including Designing Data-Intensive Applications and Alex Xu's System Design Interview series, takes you from first principles (CAP theorem, Raft consensus, Saga transactions) through production-grade building blocks (Kafka, Redis) to designing real systems end-to-end. It culminates in advanced architectural patterns, the C4 Model, geospatial proximity services, deep-dive Apache Kafka internals, comprehensive deep dives into Load Balancers and Content Delivery Networks (CDNs), and the Ultimate Tool Kit for tackling System Design Interviews.
7 chapters · Strategic & Tactical Patterns · Java
A curriculum synthesizing the foundational literature of Domain-Driven Design (Eric Evans, Vladik Khononov) to tackle complexity in the heart of software. It teaches how to bridge the gap between domain experts and developers using a Ubiquitous Language. Covers Strategic Design (Bounded Contexts, Context Mapping), Tactical Design (Entities, Value Objects, Aggregate Roots), Domain Events, CQRS with Event Sourcing, and a comprehensive visual encyclopedia of all DDD Building Blocks. All code examples are implemented in modern Java (17+).
85 chapters · 23 phases · Kernel to UNIX Power Tools
A hardcore 85-part curriculum synthesized from 34 professional-grade textbooks. Spans 23 phases: from CLI survival and Ring-0 device drivers through eBPF observability, containerization internals, FUSE filesystems, SELinux/Seccomp security, XDP/DPDK kernel-bypass networking, live kernel patching, cybersecurity operations, advanced systems programming (epoll/IPC), TCP/IP protocol mastery, extreme Performance Engineering, and culminating in digital forensics, latency tracing, and the GNU Build Ecosystem. Packed with hands-on execution exercises.