Architecting
Digital Systems

Building fast, scalable systems and clean interfaces.

About Daniel Blanco

Daniel Blanco is a full-stack software engineer specialising in scalable web systems, AI-enabled backends, and cloud-native infrastructure. He builds production-grade platforms using PHP and Laravel on the backend, Python and FastAPI for AI services, React and Next.js on the frontend, and AWS with Docker for deployment. He integrates large language models — including Anthropic Claude and the OpenAI API — into developer workflows using LangChain, pgvector, and Retrieval-Augmented Generation. His project work includes RedForge, a self-hosted adversarial testing library for LLM applications; TraceForge, an agent runtime tracing and LLM replay testing library; EvalKit, a self-hosted multi-model AI evaluation framework with hallucination detection and regression alerts; Runnory, a gamified narrative running app with an LLM biometric pipeline; and a RAG-powered codebase intelligence tool. Daniel is open to full-stack and AI-focused engineering roles. Contact: hello@danblanco.dev

Explore Works arrow_right_alt

The Toolchain

Engineering stack across backend, frontend, integrations, cloud, and machine learning.

code

Programming Languages

PHP, JavaScript (ES6+), Python, HTML5, CSS3, SQL

integration_instructions

Frameworks & Libraries

Laravel, React.js, WordPress, jQuery, Bootstrap, Tailwind CSS, FastAPI

psychology

AI & Machine Learning

LangChain, OpenAI API, Anthropic Claude API, Ollama, Voyage AI, LLM Integration, RAG, NLP, Prompt Engineering, Vector Embeddings

extension

CMS & Platform Integration

WordPress, CiviCRM, Moodle LMS, Custom Plugin Development

storage

Databases

MySQL, PostgreSQL, Redis, MongoDB, Vector Databases (pgvector)

rule

Testing & QA

PHPUnit, Jest, Cypress, Test-Driven Development (TDD), Code Review, QA Testing

build

Build & Package Tools

Webpack, Vite, Composer, NPM, Yarn

cloud

Cloud & DevOps

AWS (EC2, S3, RDS, Lambda, CloudWatch), Docker, Docker Compose, CI/CD, Jenkins

monitoring

Monitoring & Error Tracking

Sentry, AWS CloudWatch, Performance Optimization, Application Monitoring

dns

Server & Hosting

Apache, Nginx, Linux (Ubuntu, CentOS), Windows Server, High-Traffic Optimization

fork_right

Tools & Version Control

Git, GitHub, GitLab, Bitbucket, Jira, Confluence, Tree-sitter

palette

CSS Preprocessors

Sass, Less, PostCSS

hub

APIs & Integration

RESTful APIs, GraphQL, OAuth, JWT, Third-party API Integration, Webhooks

analytics

Data Science & Analytics

Python, Pandas, NumPy, Machine Learning, Data Analysis, Statistical Modeling

account_tree

Methodologies

Agile, Scrum, Test-Driven Development (TDD), Code Review, QA Testing

workspace_premium

Specialized

Blockchain Development, Payment Gateway Integration, PCI DSS Compliance, Performance Optimization, AI Automation, AST Parsing

Case Studies

Engineering projects shipped to production.

RedForge LLM — adversarial security testing report showing prompt injection and jailbreak scan results with severity dashboard
PROMPT INJECTION + JAILBREAK PYTHON + ASYNC-FIRST
01 / ADVERSARIAL LLM SECURITY

RedForge

Self-hosted adversarial testing library for LLM applications — pip-installable, async-first, framework-agnostic. Two-stage scoring with deterministic heuristic and LLM judge escalation. Nine calibrated attack variants across prompt injection and jailbreak categories. Produces self-contained HTML reports with severity dashboards, JSONL run artifacts for replay, and scan diffing for regression tracking across model versions.

Tech Stack: Python, asyncio, Anthropic / OpenAI / Ollama judges
Impact: Catches prompt injection and jailbreak regressions before deployment; diffable JSONL artifacts enable per-run comparison without re-running judges

github.com/Danultimate/redforge-llm open_in_new
02 / AGENT TRACING + REPLAY TESTING

TraceForge

Python library for tracing agent executions and replaying them without API costs. Every LLM call, tool invocation, error, and state transition is recorded as a typed span in a JSONL artifact. A first-class pytest plugin re-runs agents against those cached responses to verify execution paths without hitting the API. Generates self-contained HTML reports per run; no cloud services required.

Tech Stack: Python, asyncio, Anthropic / OpenAI / LangChain instrumentors
Impact: Replay traces without API costs; snapshot assertions catch execution path regressions before deployment

github.com/Danultimate/traceforge open_in_new
TraceForge — agent execution trace report showing LLM calls, tool invocations, and replay test results
PYTEST PLUGIN + LLM REPLAY ANTHROPIC + OPENAI
03 / AI EVALUATION FRAMEWORK

EvalKit

Self-hosted evaluation framework for running prompt test suites against Claude, GPT-4o, and Gemini simultaneously. Four scoring methods — exact match, semantic similarity via Voyage AI, Claude-as-judge, and rubric-based scoring — surface quality regressions before they reach production. Hallucination detection flags and explains each occurrence per result. Cost and latency are recorded per model and per run, with regression alerts firing automatically on score drops of 10% or more versus any saved baseline.

Tech Stack: Python, FastAPI, React 18, PostgreSQL + pgvector, Voyage AI, Anthropic Claude, OpenAI, Gemini
Impact: Multi-model evaluation in a single run; SSE streaming delivers live results without polling; regression alerts prevent silent quality drops

evalkit.danblanco.dev open_in_new
EvalKit dashboard — multi-model AI evaluation showing prompt test results across Claude, GPT-4o, and Gemini with cost, latency, and regression tracking
FASTAPI + PGVECTOR CLAUDE + VOYAGE AI
04 / GAMIFIED NARRATIVE RUNNING APP

Runnory

A real-time biometric narrative engine built into a cross-platform running app. A 5-node LangGraph pipeline processes heart rate data every 7 seconds, translating BPM and pace into live story events across three fantasy worlds. A three-float world state vector tracks tension, discovery, and energy — driving narrator responses, XP rewards, and artifact unlocks dynamically per run.

Tech Stack: React Native, Expo, LangGraph, BLE, GPS, Node.js, AI Narrative Engine
Impact: Pre-launch waitlist, iOS & Android, Strava sync integration

runnory.com open_in_new
Runnory app — world selection screen showing Shadow Realm, The Wandering Wilds, The Last Signal, Blood and Olympus, and Neon Fugitive narrative worlds
REACT NATIVE + LANGRAPH BLE + GPS
FASTAPI + PGVECTOR CLAUDE + VOYAGE AI
05 / RAG DEVELOPER TOOLING

Codebase Intelligence Engine

RAG-powered developer assistant that indexes entire GitHub repositories using Tree-sitter AST-aware chunking — splitting code at function and class boundaries rather than arbitrary token windows. Developers ask natural language questions and receive cited answers with exact file paths and line numbers, backed by Voyage AI voyage-code-3 embeddings and Claude Sonnet for synthesis.

Tech Stack: Python, FastAPI, pgvector, Voyage AI, Anthropic Claude, Tree-sitter, React, Docker Compose
Impact: Reclaims ~5 hours/week per senior engineer; reduces new hire time-to-first-commit by eliminating "where is X" questions

Download CV

Full resume with project history, technical depth, and leadership context.

description

Professional Resume

Daniel Blanco CV

Covers tech stack, project highlights, and measurable results across full-stack and AI-integrated work.

Download CV download

Talk With
My AI Assistant.

Ask about project fit, architecture decisions, and how I work. The assistant can collect the details and send me a summary.

alternate_email

Direct Email

hello@danblanco.dev

What the assistant can do

  • check_circleRecommend the most relevant case study based on your business goals.
  • check_circleAnswer technical questions about architecture, stack choices, and delivery process.
  • check_circleCollect project requirements and share a structured opportunity summary.

The assistant runs on private VPS infrastructure and only stores lead details when you explicitly provide contact information.

Frequently Asked Questions

Who is Daniel Blanco?
Daniel Blanco is a full-stack software engineer with expertise in PHP, Python, React, AWS, and AI integration, available for hire. He builds scalable web systems and integrates large language models into production workflows.
What technologies does Daniel Blanco work with?
Daniel Blanco works with Laravel, FastAPI, React.js, Next.js, AWS (EC2, S3, RDS, Lambda), Docker, LangChain, Anthropic Claude API, OpenAI API, PostgreSQL, pgvector, MySQL, Redis, WordPress, CiviCRM, and Moodle LMS.
What kind of projects has Daniel Blanco built?
Daniel Blanco has built RedForge — a self-hosted adversarial testing library for LLM applications covering prompt injection and jailbreak attacks with two-stage scoring and HTML report generation — TraceForge, a Python library for recording agent runtime traces as typed spans and replaying them via a pytest plugin without API costs — EvalKit, a self-hosted AI evaluation framework for testing prompts across Claude, GPT-4o, and Gemini with four scoring methods, hallucination detection, and regression alerts — Runnory, a gamified narrative running app with a LangGraph biometric pipeline translating heart rate into live story events — and a RAG-powered codebase intelligence tool using Tree-sitter AST chunking and Voyage AI embeddings.
How can I contact Daniel Blanco?
You can contact Daniel Blanco by email at hello@danblanco.dev, through LinkedIn at linkedin.com/in/daniel-blanco-dev/, or by using the AI assistant at danblanco.dev.
Is Daniel Blanco available for hire?
Yes. Daniel Blanco is open to full-stack and AI-focused engineering roles. He can be reached at hello@danblanco.dev.