Services · Projects · Technical Skillset · Open Knowledge Vault
Gonzalo Romero, aka 𝔇𝔢𝔢𝔭ℜ𝔞𝔱
AI engineer specialized in multimodal generative AI and applied AI systems. Passion for science since the age of 6 thanks to Carl Sagan. I love technology and information; they are the pillars of my daily life. Programming, studying and experimenting since my first computer. More than 12 years of continuous training between courses and certifications (20+), university and self-taught learning.
Services
✬ Applied AI System Development (Desarrollo de sistemas de IA aplicada) ✬ — End-to-end delivery of production-grade GenAI backends (RAG + agents): data ingestion, retrieval, orchestration, APIs, evaluation harnesses, observability, and secure deployment.
IT Consulting (Consultoría de TI) — Design and implementation of generative AI architectures, RAG systems, and agentic backends; backend/MLOps advisory; and model integration for production environments.
Cybersecurity (Ciberseguridad) — Security analysis and threat modeling for AI systems; implementation of multi-tenant RBAC, data masking, and audit logging to meet compliance requirements.
Backup & Recovery Solutions (Soluciones de copia de seguridad y recuperación) — Design of data ingestion pipelines and vector storage with backup, disaster recovery, and efficient scaling strategies.
Training (Formación) — Technical workshops and training on GenAI, RAG, agents, LangChain, vector databases, MLOps, and secure AI deployment.
Advisory & Guidance (Asesoramiento y orientación) — Strategic guidance on applied AI, model selection, system evaluation, and technology roadmap development.
Projects
Cortex — Corporate Knowledge Assistant
Enterprise-grade RAG system built with hexagonal architecture, multi-tenant RBAC, PII-aware processing, and a full audit trail. Designed for regulated environments with strict data-leak prevention and production-grade observability.
Stack: RAG · FastAPI · Qdrant · React · Redis · PostgreSQL · Docker · RBAC · DLP/PII
→ Repository: https://github.com/DeepRatAI/cortex-knowledge-assistant
MedX — Medical AI Assistant (Under Renovation)
RAG-powered assistant for clinical document Q&A, integrating LangChain orchestration with the Kimi K2 model to deliver source-grounded responses for medical information retrieval and reasoning workflows.
Stack: RAG · LangChain · Kimi K2 · Python · NLP · Vector Search
→ Repository: https://github.com/DeepRatAI/Med-X-KimiK2-RAG
CDR — Clinical Data Research · Agent Orchestration
Coming Soon — 17/02/2026
Multi-agent orchestration system for clinical data research. Autonomous agents collaborate to retrieve, validate, analyze, and synthesize clinical datasets and literature — designed for scalable, evaluation-driven research workflows in healthcare.
Stack: Agents · Orchestration · LangGraph · RAG · Python · Evaluation Harness · Observability
Technical Skillset
⯈ GenAI & Applied AI Systems
- LLMs: GPT, Mistral, LLaMA, Falcon, Kimi, Qwen-VL
- RAG: LangChain, LangGraph, FAISS, Chroma, Weaviate, Pinecone
- Vector DBs: Qdrant, Milvus, ChromaDB
- Agents: ReAct, AutoGPTQ, OpenAgents
- Fine-Tuning: LoRA, QLoRA, PEFT, bitsandbytes
- Formats: GGUF, llama.cpp
⚙ AI Modeling & Training
- Frameworks: PyTorch, Transformers, scikit-learn, XGBoost
- Workflows: training pipelines, quantization, distillation, eval metrics
- Vision: OpenCV, YOLOv8, CLIP, BLIP, ViT, Stable Diffusion, SDXL
✦ NLP & Data Intelligence
- Embeddings: SentenceTransformers, Cohere, OpenAI, HF models
- Processing: spaCy, NLTK, regex, custom pipelines
- Graph reasoning: GraphRAG, LangGraph memory
➤ Backend Engineering
- Languages: Python (Advanced)
- APIs: FastAPI, Flask, WebSocket
- Auth & Security: JWT, OAuth2, RBAC, multi-tenant
- DBs: PostgreSQL, MySQL, SQLite, MongoDB
- Observability: logging, Prometheus, Grafana
⧉ MLOps & Deployment
- Hosting: Hugging Face, Ollama, Replicate, Triton, ggml
- CI/CD: GitHub Actions, Railway, Docker Compose, Make
- Cloud: AWS, GCP
- Monitoring: LangSmith, inference logs, custom APIs
🞄 Full-Stack & UI Dev
- Frontend: Gradio, Streamlit, HTML5, CSS3, JS, TailwindCSS
- UX: prompt UIs, streaming interfaces, agents-as-apps
- Interfaces: Jinja2, Markdown rendering
Data & Analytics
- Tools: NumPy, Pandas, Seaborn, UMAP, Matplotlib
- Visualization: embedding plots, attention maps, token analysis
- Workspaces: Jupyter, Google Colab
Key Strengths
- ✔ End-to-End AI system design: from dataset to API & UI
- ✔ Multimodal & RAG-based systems in production
- ✔ Strong blend of backend engineering, ML, and DevOps
- ✔ Obsessed with tooling, benchmarks, and real-world reliability
⚙ Open Knowledge Vault — Certifications & Specializations ⚙
Click to expand full list
- IBM AI & Machine Learning Professional Certificate
- IBM Generative AI Foundations
- Mathematics for Machine Learning — Duke University
- Deep Learning — IBM
- Advanced Machine Learning and Signal Processing — IBM
- Intro to Computer Vision and Image Processing — IBM
- Python for Data Science, AI & Development — IBM
- Databases and SQL for Data Science — IBM
- Tools for Data Science — IBM
- Data Visualization with Python — IBM
- Data Analysis with Python — IBM
- Machine Learning with Python (with Honors) — IBM
- Deep Neural Networks with PyTorch — IBM
- Deep Learning with TensorFlow — IBM
- Machine Learning with Python — IBM Developer Skills Network
Full record available on LinkedIn: https://www.linkedin.com/in/gonzalo-romero-b9b5b4355/