Projects and professional work — Alejandro Newport

Client work, teaching, products and applied AI: the things built, what was contributed to each, and what can be verified about them.

Who he is, what he studied and how to reach him are on the profile page.

Professional experience

• AI & Data Analysis Engineer — Metaview

  Jun 2026 – Sep 2026

  Built and evaluated retrieval, vision and analytics systems over large document, image and record corpora.

  • Processed and structured a corpus of over 500,000 documents for retrieval and generation.
  • Prototyped a RAG chatbot answering internal queries against the corpus.
  • Trained and evaluated an image classifier over a 50,000+ image dataset.
  • Built automated Python and SQL pipelines for ingestion, transformation and reporting.
  • Evaluated generation quality with BLEU, ROUGE and perplexity.
  • Documents processed: 500K+
  • Text-generation relevance: +35%
  • RAG query accuracy: 90%+
  • Support response time: −60%
  • Images classified: 50K+
  • Image classification accuracy: 92%
  • Manual labelling time: −70%
  • Records analysed: 1M+
  • Reporting time: −40%
  • Churn forecast accuracy: +25%
  • Coherence score: +25%

  Technologies: Python, SQL, NLP / RAG, Deep Learning, TensorFlow, Pandas, NumPy, Model Evaluation, Feature Engineering, EDA.

• Graduate Teaching Assistant — King's College London

  Sep 2025 – Jun 2026

  Ran labs and tutorials, debugged student code live, and taught testing, complexity and clean code.

  • Led weekly labs and tutorials for undergraduate computer science modules.
  • Debugged student code live and taught systematic fault isolation.
  • Taught testing discipline, algorithmic complexity and clean code practice.

  Technologies: Java, Python, Testing, Algorithms, Debugging.

• Full-Stack Web Developer — Pansofia / Grupo Newport

  Jun 2025 – Sep 2025

  Led a team of six engineers and designers delivering fourteen production websites, two weeks ahead of schedule.

  • Led a team of 6 software engineers and designers.
  • Delivered 14 professional websites into production.
  • Completed launches two weeks ahead of schedule.
  • Personally built the majority of the front-end and back-end across the portfolio.
  • Deployed and maintained the sites on private server infrastructure.
  • Delivered measurable increases in client engagement.
  • Websites shipped: 14
  • Engineers & designers led: 6
  • Ahead of schedule: 2 weeks
  • Front & back-end built: 80%+

  Technologies: JavaScript, TypeScript, Vue.js, React, Laravel, PHP, MySQL, HTML, CSS, Git, Linux.

• Co-Founder & Developer — Focus

  May 2024 – Jun 2026

  AI-powered educational application that generates personalised short-form learning videos end to end.

  • Co-founded and built an AI application generating personalised short-form learning videos.
  • Designed the pipeline from prompt to script to narrated, captioned, rendered video.
  • Integrated content generation, voice synthesis, captioning and multiple visual formats.
  • Built the mobile experience on top of automated rendering pipelines.

  Technologies: Python, TypeScript, React, NLP, Deep Learning, Video Rendering, Text-to-Speech, Mobile.

Project archive

• Focus

  2024 · personal · AI product

  An educational application that turns a topic into a finished, narrated, captioned short-form video. A generation pipeline writes the script, synthesises the voice, times the captions, selects the visual format and renders the result — then serves it inside a mobile experience built around the pipeline rather than bolted onto it.

  Contribution: Co-founder and developer. Designed and built the generation pipeline and the application around it.

  • Co-founded and built an AI application generating personalised short-form learning videos.
  • Pipeline runs prompt → script → voice → captions → visual composition → rendered video.
  • Supports multiple visual formats over a single content model.
  • Mobile experience sits on top of automated rendering pipelines.

  Technologies: Python, TypeScript, React, NLP, Deep Learning, Text-to-Speech, Video Rendering, Mobile.

• Gym App

  2026 · personal · Rigged animation system

  An exercise system built the expensive way once instead of the cheap way a thousand times: a single rigged 3D athlete, a library of pose and motion data, and a set of adapted gym equipment. Every exercise is a data record that drives the same model, rather than a separate asset — so the library scales without the download size scaling with it.

  Contribution: Concept, 3D pipeline, pose-data architecture and application.

  • Architecture: one reusable rigged model + pose data + equipment definitions, composed at runtime.
  • The 3D asset is loaded once and re-posed, never duplicated per exercise.

  Base models loaded: 1. Exercises: configurable.

  Technologies: 3D, Rigging, Animation Systems, Swift, Data Modelling, Mobile.

  Gym App — source repository

• Chess Assistant

  2025 · personal · Computer vision

  A computer-vision pipeline that reads a chess position from pixels. One network regresses the board corners out of a full screenshot; a second classifies each of the 64 squares into thirteen classes. The reconstructed position is serialised to FEN and handed to Stockfish, which returns the best move. Training data is synthetic — boards composited onto generated desktops at random positions, sizes and themes, so the detector generalises past any one client.

  Contribution: Full pipeline: dataset synthesis, both models, board reconstruction, engine integration and deployment.

  • Two trained models are committed to the repository: a per-square piece classifier and a board-corner regressor.
  • Piece classifier is a 3-block CNN (Conv2D 32/64/128 with max-pooling) → Flatten → Dense(128) → Dropout, on 64×64 crops.
  • Board localisation is solved as direct corner regression on a MobileNetV2 backbone at 384×384, refined by a local search.
  • Training data is synthetic: boards composited onto generated desktop wallpapers at random position, scale and theme.
  • Reconstructed positions are serialised to FEN and analysed by Stockfish via python-chess.
  • Thirteen piece classes (six white, six black, empty).

  Piece recognition accuracy: 99.6%. Synthetic boards generated: 1,200. Piece classes: 13. Board themes simulated: 5. Detector input: 384×384. Classifier input: 64×64.

  Technologies: Python, TensorFlow, Keras, MobileNetV2, OpenCV, NumPy, python-chess, Stockfish, AWS EC2, AWS S3, Deep Learning.

  Chess Assistant — source repository

• Cinquillo 2.0

  2026 · university · Game AI & experiment design

  A research project that treats game design as an optimisation problem. A fully parameterised Cinquillo engine exposes 33 named rule variants; Random, Heuristic, MCTS and reinforcement-learning agents play them at scale; and a 16-experiment pipeline measures how much of each outcome is skill and how much is seat order and deal luck. The output is not a bot — it is evidence for which rules make the game fair.

  Contribution: Engine, agents, experiment harness and analysis.

  • Engine models the 40-card Spanish deck (Oros, Copas, Espadas, Bastos; ranks 1–7 and 10–12) with legality derived from board adjacency.
  • Moves are an abstract base class with PlayCard / RollDice / Pass subclasses — variants compose rather than fork.
  • 33 named rule variants, including a fully parameterised dice mechanic with four beneficial and several adverse effects.
  • Agents span Random, Heuristic, Monte Carlo Tree Search and a reinforcement-learning agent.
  • RL agent uses a 209-feature state encoding, a 42-action Q-network and a 100,000-transition replay buffer.
  • 16 simulation experiments, 5,000 games each by default.
  • README reports 260 passing tests.

  Rule variants: 33. Simulation experiments: 16. Games per experiment: 5,000. RL state features: 209. Q-network actions: 42. Replay buffer: 100K. Tests passing: 260.

  Technologies: Python, NumPy, SciPy, pandas, matplotlib, seaborn, Flask, pytest, Reinforcement Learning, MCTS.

  Cinquillo 2.0 — source repository

• Stock Market Simulator

  2025 · personal · Concurrency

  Trader threads place market and limit orders against independent per-symbol order books. A matching engine enforces price-time priority, executes crossing orders at the resting order’s limit price, re-rests unfilled remainder, and tracks position and PnL — all while a price engine walks each symbol under a Gaussian random walk and a JavaFX dashboard renders it live.

  Contribution: Whole system: order books, matching engine, price engine, threading model and dashboard.

  • Order book implements price-time priority with two priority queues and explicit comparators.
  • Matching engine executes crossing limit orders at the resting order’s limit price and re-rests the remainder.
  • Price engine applies a geometric random walk per tick (≈0.1% standard deviation) at 5 ticks per second.
  • Concurrency via java.util.concurrent with a ScheduledExecutorService driving ticks.
  • Live JavaFX dashboard renders books, fills and PnL as the simulation runs.
  • Pluggable trader strategies, each running on its own thread.

  Concurrent trades handled: 40,000+. Price ticks: 5/sec. Trader threads: 4. Symbols: 3. Starting cash per trader: 100,000.

  Technologies: Java, JavaFX, Maven, Concurrency, Multithreading.

  Stock Market Simulator — source repository

• Keyframes

  2026 · personal · Full-stack platform

  A full-stack catalogue of motion-graphics work: a filterable public grid with hover previews and detail pages, a public suggestion form, and a private back-office where an editor approves a suggestion into a live record. The interesting parts are defensive — an allow-listed embed builder, a four-gate upload validator, and accent-insensitive search that works identically on SQLite and Postgres.

  Contribution: Sole author. Monorepo, API, database design, front-end and admin.

  • npm-workspaces monorepo: an Express + Prisma API and a React 19 + Vite front end sharing type, constant and embed modules.
  • Six Prisma models, with parallel SQLite and PostgreSQL schemas kept behaviourally identical (no native enums, JSON columns or scalar lists).
  • Accent- and case-insensitive search without Postgres-only features: rows store a pre-normalised search haystack and queries normalise the needle identically.
  • Video embeds never take an editor-supplied URL: the host is matched against an allow-list, the id validated per provider, and the embed URL rebuilt from a fixed template.
  • Uploads pass four independent gates — declared MIME, matching extension, byte size, and a real magic-byte signature check after write.
  • Helmet CSP configured with useDefaults: false and an explicit directive set; the uploads directory gets its own tighter policy.
  • Auth is a JWT in an httpOnly SameSite cookie with bcrypt hashes; the seed script creates no user and there are no default credentials in the repo.
  • Editorial pages are database rows rendered as React elements from Markdown, so page copy cannot inject markup.

  Prisma models: 6. Upload validation gates: 4. Video providers supported: 4. Seeded works: 66. Countries in seed data: 28. Route modules: 15.

  Technologies: TypeScript, React, Vite, Node.js, Express, Prisma, PostgreSQL, SQLite, Zod, JWT.

  Keyframes — source repository

• TappedIn

  2025 · university · NLP + web application

  A job board where the matching is the product. Uploaded CVs are parsed with spaCy and PDF extraction into structured skills; a weighted matcher scores each candidate against each role out of 100, separating required from preferred skills and reporting exactly which required skills are missing. Built by a team of six, with a test suite to match.

  Contribution: Team of six. Contributed across the matching engine, CV parsing and application.

  • Match score is 0–100 and returns both matched skills and the list of missing required skills.
  • Required and preferred skills are weighted 70/30, with a contract-type match bonus.
  • CV parsing uses pdfplumber for text, PyMuPDF for validation and spaCy en_core_web_sm for entity extraction.
  • The matcher carries a hand-built skill-synonym table (js/javascript, k8s/kubernetes, ml/ai, ux/ui …).
  • 333 test methods across 30 test files.
  • Seeded with 100 users, 200 jobs and 500 applications.

  Test methods: 333. Skill weighting: 70 / 30. Match score range: 0–100. Team size: 6.

  Technologies: Python, Django, spaCy, pdfplumber, PyMuPDF, PostgreSQL, SQLite, Docker, Bootstrap, Selenium, NLP.

  TappedIn — source repository

• Orca

  2024 · university · Scheduling system

  A tutoring scheduling platform with three roles, recurring lesson generation, timetable clash detection and invoicing. The scheduling logic is the hard part: a request expands into a recurrence series, every instance is checked against both tutor and student availability, and conflicts surface before anything is booked.

  Contribution: Team of five. Contributed to scheduling, models and test coverage.

  • Custom user model extending AbstractUser with an admin/tutor/student type and a username regex validator.
  • Five models: Subject, User, Lesson, Invoice, Notification.
  • 366 test methods across 53 test files.
  • 37 database migrations.
  • Lessons range 30–240 minutes inside an 08:00–20:00 booking window.
  • Seeded with 500 users.

  Test methods: 366. Migrations: 37. Seeded users: 500. Team size: 5.

  Technologies: Python, Django, SQLite, Bootstrap, Faker, Selenium, pytest.

  Orca — source repository

• Three Body Problem

  2025 · personal · Physics simulation

  A three-body gravity simulator built on a velocity-Verlet integrator with gravitational softening, six named orbital presets, and a live total-energy readout that shows the integrator conserving energy — or failing to, when you push it. Zero dependencies.

  Contribution: Sole author.

  • Integrator is velocity-Verlet: accelerations computed, positions advanced by v·dt + ½a·dt², accelerations recomputed, velocities corrected.
  • Gravitational softening avoids singular forces at close approach.
  • Pairwise force accumulation is O(n²) exploiting Newton’s third law reciprocity.
  • Six orbital presets, including the figure-eight solution.
  • Trails capped at 2,000 points; no runtime dependencies.

  Orbital presets: 6. Max trail length: 2,000. Dependencies: 0.

  Technologies: JavaScript, Canvas 2D, Numerical Methods, Physics.

  Three Body Problem — source repository

  Three Body Problem — live site

• Labyrinth

  2025 · personal · Algorithms

  Three maze generators and eight search algorithms, all animated on the same grid so you can watch how differently they think. Recursive backtracker, Prim and Kruskal build; BFS, DFS, Dijkstra, A*, greedy best-first and others solve — and the difference between a breadth-first wavefront and an A* teardrop is immediately legible.

  Contribution: Sole author.

  • Eight search algorithms and three maze generators.
  • Generators: iterative-DFS recursive backtracker, Prim, and Kruskal.
  • Kruskal is backed by a union-find structure with path compression and union by rank.
  • Grids up to 120×120. No JavaScript libraries.

  Search algorithms: 8. Maze generators: 3. Max grid: 120×120. Libraries: 0.

  Technologies: JavaScript, Canvas 2D, Algorithms, Data Structures.

  Labyrinth — source repository

  Labyrinth — live site

• Encrypted Proxy VPN

  2025 · personal · Networking

  A working, minimal VPN: a threaded TCP relay on a home machine, a local HTTP proxy on the remote device, and Fernet symmetric encryption on the wire between them. Plus a browser-side detector page that measures whether traffic is actually going through the tunnel.

  Contribution: Sole author. Server, client proxy, crypto layer and detector page.

  • Server binds a TCP socket with SO_REUSEADDR and spawns a daemon thread per connection.
  • Payloads are encrypted with Fernet (symmetric authenticated encryption) from the cryptography library.
  • Client runs a local HTTP proxy that forwards requests over the encrypted relay.
  • Includes a browser detector page that measures latency and connection routing.
  • No third-party dependency manifest — dependencies are declared in module docstrings.

  Relay port: 8080. Local proxy port: 8888. Response buffer: 64 KB.

  Technologies: Python, Sockets, Threading, Cryptography, Networking, JavaScript.

  Encrypted Proxy VPN — source repository

  Encrypted Proxy VPN — live site

• Minecraft Seed Finder

  2025 · personal · Procedural generation

  A brute-force seed search over Minecraft’s world generation. Python drives four generated C programs built against the Cubiomes library, scanning ten thousand regions per seed to find worlds where a stronghold sits within 500 blocks of a required set of structures.

  Contribution: Sole author.

  • Python writes four complete C programs into a Cubiomes checkout, compiles them, and drives them as subprocesses.
  • All four C programs target Minecraft 1.21 world generation.
  • Scans 10,201 overworld regions per query and 90,601 nether regions for fortress queries.
  • Structure distance limit of 500 blocks around each of three checked strongholds.

  Regions scanned per query: 10,201. Nether regions per fortress query: 90,601. Structure distance limit: 500 blocks. Minecraft version: 1.21.

  Technologies: Python, C, Cubiomes, matplotlib, Make, Procedural Generation.

  Minecraft Seed Finder — source repository

• ML Revision Engine

  2026 · university · Learning tool

  A whole revision system in one 182 KB HTML file — 188 questions across 14 topics, 83 flashcards, 39 formula cards and eight quiz modes, with progress persisted to local storage and 25 questions explicitly flagged as traps. No build step, no dependencies.

  Contribution: Sole author.

  • Entire project is one index.html of 181,828 bytes / 3,798 lines.
  • 188 questions across 14 topics, 83 flashcards, 39 formula cards.
  • 25 questions flagged as hard/trap questions.
  • Eight quiz modes; progress persisted with the Web Storage API.
  • Zero dependencies and no build step.

  Questions: 188. Flashcards: 83. Topics: 14. Quiz modes: 8.

  Technologies: JavaScript, HTML, CSS, Web Storage API.

  ML Revision Engine — source repository

  ML Revision Engine — live site

• Robot Arm Portfolio

  2026 · personal · WebGL

  An earlier portfolio experiment: a rigged robot-arm GLB driven by inverse-kinematics-style targeting toward the cursor, lit by image-based lighting and a five-light rig, with an idle sleep pose after inactivity.

  Contribution: Sole author.

  • All scene logic in a single ~20 KB file — build, animate and teardown in one effect.
  • Image-based lighting via PMREMGenerator + RoomEnvironment plus a five-light rig with a 4096² shadow map.
  • Seven rig nodes driven per frame; the arm returns to a sleep pose after 3.8 s idle.
  • ACES filmic tone mapping, high-performance WebGL context.

  Rig nodes driven: 7. Shadow map: 4096². Idle sleep threshold: 3.8 s.

  Technologies: React, Three.js, WebGL, Vite, glTF.

  Robot Arm Portfolio — source repository

• Personal Site (2024)

  2024 · personal · web

  A hand-written static site — three files, no build step — with English/Spanish switching, dark and light themes, five responsive breakpoints and complete Schema.org, Open Graph and JSON-LD markup.

  Contribution: Sole author.

  • No package manifest and no build step — three hand-written files.
  • Two languages, five responsive breakpoints, thirteen project cards.
  • Schema.org JSON-LD, Open Graph and Twitter metadata.

  Project cards: 13. Languages: 2. Breakpoints: 5.

  Technologies: HTML, CSS, JavaScript, SEO.

  Personal Site (2024) — source repository

  Personal Site (2024) — live site

• My Library

  2024 · university · web

  A small, correct Django application: one model, a strict ISBN validator accepting only 10- or 13-digit values, Faker-driven seeding, and 55 unit tests running in GitHub Actions on every push.

  Contribution: Sole author.

  • Django 5.1.2 project with a single app; one Book model.
  • ISBN field is unique and validated by a regex accepting exactly 10 or 13 digits.
  • 55 unit tests, run in GitHub Actions on Python 3.12.
  • Seed command generates 100 books via Faker.

  Unit tests: 55. Seeded books: 100.

  Technologies: Python, Django, SQLite, Bootstrap, GitHub Actions, Faker.

  My Library — source repository

• VPN Profile Generator

  2025 · personal · software

  A single page that builds valid Apple configuration profiles entirely client-side — no server ever sees the credentials. Emits correct plist payloads for three VPN protocols and hands back a downloadable .mobileconfig.

  Contribution: Sole author.

  • Entire project is one 907-line index.html with no build step or dependencies.
  • Generates valid Apple .mobileconfig plists for IKEv2, L2TP/IPSec and OpenVPN.
  • IKEv2 payload sets AES-256 / SHA2-256 / DH group 14.
  • Everything runs client-side; the file is delivered via a Blob object URL.

  Protocols: 3. Dependencies: 0.

  Technologies: JavaScript, HTML, Apple Configuration Profiles, Networking.

  VPN Profile Generator — source repository

  VPN Profile Generator — live site

• Cinquillo

  2025 · personal · experiment

  The Spanish card game in one 15.7 KB HTML file, with bot opponents and the full adjacency rule set. The precursor to the fairness research project.

  Contribution: Sole author.

  • Entire project is one index.html of 15,725 bytes.
  • Models the 40-card Spanish deck (four suits × ranks 1–7, 10–12).
  • Two to four players; zero dependencies.

  Deck: 40 cards. Source: 1 file, 15.7 KB.

  Technologies: JavaScript, HTML, CSS.

  Cinquillo — source repository

  Cinquillo — live site

• Dots & Boxes: Dice Quotas

  2025 · personal · experiment

  The classic game with one rule change that transforms it: each turn you roll a die and must draw exactly that many edges, clamped to what remains. Suddenly the chain strategy inverts.

  Contribution: Sole author.

  • Single index.html of 569 lines, zero dependencies.
  • Board sizes 3×3 to 6×6.
  • A die roll sets the turn quota, clamped to the number of edges remaining.

  Board sizes: 3×3 – 6×6. Source: 569 lines.

  Technologies: JavaScript, Canvas 2D, HTML.

  Dots & Boxes: Dice Quotas — source repository

  Dots & Boxes: Dice Quotas — live site

• Primos en Click

  2024 · personal · experiment

  Seventy-one lines of code, total. Included here because small things that do exactly one thing are worth keeping.

  Contribution: Sole author.

  • Four files, 71 lines of code in total. No dependencies, no build step.

  Lines of code: 71.

  Technologies: JavaScript, HTML, CSS.

  Primos en Click — source repository

  Primos en Click — live site

• PokeAppVue

  2024 · personal · web

  A Vue 3 Composition API app that loads a thousand records in one burst and then does all search, type filtering and paging on the client.

  Contribution: Sole author.

  • Fetches 1,000 entries on load from the public PokéAPI, then filters entirely client-side.
  • Vue 3 Composition API with script setup throughout; three components.
  • 100 cards per page.

  Records loaded: 1,000. Cards per page: 100.

  Technologies: Vue.js, Vite, Tailwind CSS, axios, JavaScript.

  PokeAppVue — source repository

• Fuerteventura 2000 (Vue prototype)

  2024 · personal · web

  An early single-page prototype for a client site that shipped for real the following summer — six routes, eighteen components, Composition API throughout.

  Contribution: Sole author.

  • Six routes registered with createWebHistory; eighteen .vue components.
  • Vue 3 Composition API with script setup throughout.
  • Tailwind CSS with PostCSS and Autoprefixer.

  Routes: 6. Components: 18.

  Technologies: Vue.js, Vite, Tailwind CSS, JavaScript.

  Fuerteventura 2000 (Vue prototype) — source repository

• Fullscreen Video Player

  2025 · personal · web

  A deliberately tiny piece of engineering: 441 bytes of JavaScript that makes a looping video autoplay reliably across browsers, including the ones that refuse until you ask correctly.

  Contribution: Sole author.

  • No package manifest and no dependencies.
  • 441 bytes of JavaScript in total.
  • Sets playsInline and handles the autoplay promise rejection path.

  JavaScript: 441 bytes.

  Technologies: JavaScript, HTML5 Video, CSS.

  Fullscreen Video Player — source repository

  Fullscreen Video Player — live site

• Catan AI

  2025 · personal · ai-ml

  A declared next project: multiple agent types for Settlers of Catan, a game whose non-determinism and trading make it substantially harder than the perfect-information games above. Listed honestly — the repository is empty.

  • Repository exists with a single commit and no source code.

  Technologies: .

  Catan AI — source repository

• Events App

  2025 · personal · mobile

  An Expo Router scaffold, kept in the archive for completeness rather than dressed up as more than it is.

  • Single commit; unmodified create-expo-app tabs template.

  Technologies: React Native, Expo, TypeScript.

  Events App — source repository

• Vue Web

  2024 · personal · web

  A learning scaffold. The one non-trivial thing in it — route-level code splitting — is what the motion graphic shows.

  • Stock create-vue scaffold with two routes and exactly two runtime dependencies.

  Routes: 2.

  Technologies: Vue.js, Vite, JavaScript.

  Vue Web — source repository

• Fuerteventura 2000

  2025 · client · Vocational training

  Subsidised employment training centre. Course catalogue, enrolment and centre information.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #19248b, #4f6d7a, #c0d6df.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Fuerteventura 2000 — live site

• Pansofía

  2025 · client · Education technology

  The group’s own agency brand: training management, e-learning and editorial services.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #111113, #e94a2f, #a2a2ab.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Pansofía — live site

• Clínica Bioever

  2025 · client · Longevity and aesthetics clinic

  Longevity and aesthetic medicine clinic — programmes, services and booking.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #0c3e46, #3fbfa8, #1c8c82.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Clínica Bioever — live site

• CHT Canarias

  2025 · client · Hospitality & tourism school

  Centro de Hostelería y Turismo de Canarias — programme listings and school information.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #15284c, #254989, #adadad.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  CHT Canarias — live site

• Hotel Escuela El Mirador

  2025 · client · Hospitality training hotel

  A working hotel that doubles as a training school — rooms, restaurant and training programmes.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #c48c3d, #3c3c3b, #333333.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Hotel Escuela El Mirador — live site

• Newport Media Films

  2025 · client · Audiovisual production

  Production company and audiovisual/drone training — showreel-led, media-heavy layout.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #d9000d, #e13510, #2e2e2d.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Newport Media Films — live site

• Escuela de Hostelería Canaria

  2025 · client · Catering school

  Vocational catering and hospitality training.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #021d30, #ff4130, #f2f3f5.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Escuela de Hostelería Canaria — live site

• ESENFUER

  2025 · client · Further education academy

  Escuela Superior de Enseñanza de Fuerteventura — private training academy.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #41b1c3, #007ebf, #00a19a.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  ESENFUER — live site

• FPE Europea

  2025 · client · Employment training & consulting

  Vocational training for employment, job placement and business consulting.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #008acf, #74c4cc, #fcc72b.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  FPE Europea — live site

• NFORMAR

  2025 · client · Employment training

  Subsidised employment training provider.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #e5222e, #dc2626, #7c7c7c.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  NFORMAR — live site

• Aula Impulsa

  2025 · client · Employment training

  Vocational and employment training courses.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #071d34, #8ede6b, #23b3a6.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Aula Impulsa — live site

• Talento Profesional

  2025 · client · Vocational training

  Training centre operating across two islands.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #d4e414, #4d5300, #1a1a1a.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Talento Profesional — live site

• Eduprisma

  2025 · client · Certified training & e-learning

  Certificados de profesionalidad and teleformación (distance learning).

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #4d2779, #f99c00, #1d1d1b.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Eduprisma — live site

• Level Up Canarias

  2025 · client · Fitness & recovery

  Personal training and recovery centre — the outlier in an education-heavy portfolio.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #9a5a5c, #854b4d, #1f1f1f.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Level Up Canarias — live site

• AvanzaFP

  2025 · client · Vocational training

  Formación profesional para el empleo.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #1230e6, #01093a, #0d1b4c.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  AvanzaFP — live site

• Innova Urbis

  2025 · client · Real estate

  Property sales and investment — multilingual, listing-driven site.

  Contribution: Built as part of the site programme delivered by a team of six, led by Alejandro. Front-end build, back-end integration and deployment to private server infrastructure.

  • Live and publicly reachable — verified by HTTP request and page content.
  • Brand palette extracted from the live public stylesheet: #ce191c, #a51418, #1c1917.
  • Source repository is private; no code, infrastructure or client data is exposed here.

  Technologies: JavaScript, HTML, CSS, PHP, MySQL, Responsive Design.

  Innova Urbis — live site