I build websites, APIs and Python automations that organize data, simplify processes and cut down repetitive work.
I'm Luis Pimentel, a Python developer in Potosí, Bolivia, open to remote freelance work.
The code of my projects is public on GitHub, with automated tests you can run yourself.
Two kinds of work, each in its own place: websites, and Python tools that handle data. Everything here has public code, and every feature, decision and test count comes from its repository, not from memory.
Sites for a person, a business or a product: the design, the content and the publishing. Here are my own site and a concept demo for each website package.
My bilingual portfolio: it presents my projects and services and makes it easy to get in touch.
One concept demo per package, each a separate site for a fictional business. A demo appears here once it is built and published.
Essential
Café
Concept demo: a fictional business
A one-page website for a fictional café in Potosí: fast, clear and easy to use on a phone, with everything a customer looks for before visiting.
Professional
Boutique hotel
Concept demo: a fictional business
The website of a fictional boutique hotel on the Atlantic: multi-page editorial design with scroll-linked animation, room galleries and a demo booking flow.
Signature
Contemporary silversmith's workshop, inspired by Potosí
Future project Not built yet: there is no demo to open. Its preview and link will appear here once it is published.
Personal projects, not client work: tools that import, check and move data. Each one runs on your own machine, and its code is public on GitHub.
Import quotes from CSV, Excel and PDF, with human review
Useful if your business receives quotes, price lists or orders as CSV, Excel or PDF files, and someone types them into a system by hand.
Personal project with public code. It runs on your own machine; there is no public demo. Its interface is in Spanish.
Quotes arrive as CSV, Excel and PDF files. Typing them into a system by hand is slow, and a wrong value or a repeated row is easy to miss and hard to undo once it is saved.
QuoteFlow imports quotes from CSV, XLSX and PDF files. Each file is parsed in an isolated process, validated and checked for duplicates, then held for review. Nothing is saved as a record until a person confirms it.
Python, FastAPI, SQLAlchemy 2, Alembic, psycopg 3, Argon2, PyJWT, openpyxl, pypdf, PostgreSQL 18, Next.js 16, React 19, TypeScript, Docker Compose, pytest, node:test, Ruff, ESLint
Nothing is stored without approval.
Rows wait in a temporary table. Confirming saves the accepted rows of the chosen sheet and deletes every temporary row of the batch in one transaction; any error rolls back.
Duplicates are private.
Rows are compared only with records confirmed by the same user. Other users' records are never compared or revealed.
Invalid rows cannot be accepted.
A request that tries to accept an invalid or empty row is rejected whole, with a 422.
Uploads never touch the disk.
The file is read in memory with a size limit and hashed with SHA-256; it is not written to disk or stored in the database.
Production mode refuses unsafe settings.
With QF_ENV=production the API will not start without secure cookies, HTTPS origins and fresh secrets.
Requirements: Docker with Compose, Python 3.12 or later (developed and tested on CPython 3.14), Node.js 24 and npm.
A fresh install on Windows PowerShell, as the project's README gives it. The README also covers an existing database, macOS and Linux.
Settings: copy the example, then replace every “change-me” value
Copy-Item .env.example .envCheck that no database exists yet: this command must fail with “no such volume”
docker volume inspect quoteflow_pgdataStart PostgreSQL, with the main and the test database
docker compose up -dBackend: virtual environment and the exact versions the tests passed with
cd backend
py -3.14 -m venv .venv
.\.venv\Scripts\python -m pip install -e ".[dev]" -c constraints.txtDatabases: the test one, then the main one (“alembic current” must print no revision)
.\.venv\Scripts\alembic -x db=test upgrade head
.\.venv\Scripts\alembic current
.\.venv\Scripts\alembic upgrade headDevelopment admin user, then the API on port 8000
.\.venv\Scripts\python -m app.cli create-admin
.\.venv\Scripts\uvicorn app.main:create_app --factory --host 127.0.0.1 --port 8000Frontend, in a second terminal; then open http://localhost:3000
cd frontend
npm ci
npm run dev512
backend tests
pytest, against a separate test database that is emptied between tests
46
frontend tests
node:test, against the production build with a fake API
Also checked with Ruff (lint and format), ESLint and the TypeScript compiler.
cd backend
.\.venv\Scripts\python -m pytest
.\.venv\Scripts\ruff check .
.\.venv\Scripts\ruff format --check .
cd ..\frontend
npm run lint
npm run build
npx tsc --noEmit
npm testLead capture form and a small CRM
Useful if enquiries reach you through a form on your website and you need to follow up each one, without duplicates and without losing track.
Personal project with public code. It runs on your own machine and has no user accounts, so it is meant for local use.
Enquiries from a website form tend to end up in an inbox or a spreadsheet. The same person can be saved twice, and it becomes hard to see who was contacted, what was said and what comes next.
LeadFlow captures leads from a web form, stores them in PostgreSQL without duplicates, and tracks each lead's status, follow-up notes and history in a simple CRM panel.
The whole flow in 13 seconds
Register a lead, open it in the CRM, change its status and add a note. Fictitious data, with the API simulated in the browser.
Fictitious data. Captured from the production build of the frontend, with the API simulated in the browser the same way the project's end-to-end tests do it.
Python, FastAPI, Pydantic 2, SQLAlchemy 2, psycopg 3, PostgreSQL 18, Next.js 16, React 19, TypeScript, Docker Compose, pytest, Playwright, Ruff
The browser never calls the API directly.
Next.js server routes forward each request to FastAPI with a timeout and return generic messages, so internal details stay hidden.
The proxy only accepts JSON.
Other content types get a 415 and bodies over 64 KB get a 413, so another site open in the browser cannot post forms to it.
Logs carry identifiers only.
event_id and lead_id, never names, emails or the text of notes.
Requirements: Node.js 20 or later (tested with 24), Python 3.11 or later (tested with 3.14) and PostgreSQL 18, with Docker Desktop or a local install.
Windows PowerShell, as the project's README gives it.
Settings: both files are ignored by Git; the example values work as they are with Docker
Copy-Item .env.example .env
Copy-Item frontend/.env.example frontend/.env.localDatabase: wait until it shows “healthy”
docker compose up -d postgres
docker compose psBackend, terminal 1; the API documentation is at http://localhost:8000/docs
cd backend
py -3.14 -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
python -m app.init_db
uvicorn app.main:app --reloadFrontend, terminal 2: the form at http://localhost:3000 and the CRM at /crm
cd frontend
npm install
npm run dev148
backend tests
pytest, against a separate test database
17
end-to-end tests
Playwright in Chromium, with the API simulated in the browser
Code checked with Ruff.
cd backend
pytest
ruff check .
cd frontend
npx playwright install chromium
npm run test:e2ePython web scraper with Excel and CSV export
Useful if you need the data from a public online catalogue in a spreadsheet, without copying it page by page.
Personal project with public code. It runs on your own machine; its tests run on GitHub Actions.
Copying product data from an online catalog, page after page, into a spreadsheet is slow, and it leaves inconsistent prices, missing fields and repeated rows.
Extracts title, price, currency, availability and URL from a paginated book catalog with Playwright, cleans the data, removes duplicates and exports Excel and CSV files. Built for scraping-practice sites such as books.toscrape.com.
python main.py --limit 50
python main.py --profile my_site.json --limit 20Python, Playwright, openpyxl, pytest, GitHub Actions
Requirements: Python 3.11, 3.12 or 3.13, and an internet connection.
Windows PowerShell, as the project's README gives it; the README also covers macOS and Linux.
Virtual environment, dependencies and the Chromium browser for Playwright
py -3.13 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -r requirements.txt
python -m playwright install chromiumCollect the first 50 books
python main.py --limit 50Another catalog, with your own site profile
python main.py --profile my_site.json --limit 20python -m pip install -r requirements-dev.txt
python -m pytestWorks on catalogs whose products are in the page HTML and paginate with links. No logins, captchas or infinite scroll.
Two kinds of work: websites in three packages, and Python tools that handle data. Prices depend on the scope, so each project starts with a short conversation.
Three packages that differ in scope, design and interaction, not only in size. None of them needs a backend.
| Essential | Professional | Signature | |
|---|---|---|---|
| Best for | A café, a shop or a professional who needs a clear presence and an easy way to be contacted | A hotel, a clinic or a business with several services that need their own space | A brand that wants its website to be part of how people remember it |
| Pages and sections | One page with up to six sections | Up to five pages, or one long page with up to ten sections | One main experience page, plus up to four supporting pages |
| Design | A clean layout adapted to your brand colours, fonts and photos | A custom design for your brand, with its own components | Original art direction: a visual concept created for your brand |
| Interactions | Links to call, write or find you on a map | An image gallery with a full-screen viewer, and pages for each service | Interactions designed for your content, such as a story told as the page scrolls |
| Animation | Subtle entrances as the page scrolls | Coordinated scroll and section animations | Elaborate scroll-driven animation and custom micro-interactions, with a reduced-motion version |
| Contact | A link to your email or WhatsApp | A contact form; an external form service, if you want one, is a separate cost | As in Professional |
| Price | Based on project scope | Based on project scope | Based on project scope |
| I want a site like this: Essential | I want a site like this: Professional | I want a site like this: Signature |
Every package includes a responsive design, basic SEO, accessibility and a site ready to publish as static files.
Not included unless agreed separately: a backend, an online shop, a content management system or user accounts.
Freelance work in the same areas as my Python projects: tools for your data and your process.
Upload CSV, Excel or PDF files, validate every row, flag duplicates and let a person approve the result before it is saved.
Collect specific fields from pages you are allowed to scrape and deliver clean, deduplicated spreadsheets.
FastAPI and PostgreSQL backends with input validation and duplicate protection, connected to a Next.js interface.
Turn repeated manual steps into a script: exact number parsing, normalized values and formatted Excel output.
Four steps, from your first message to the delivery.
You tell me what takes your team time, or what your website needs to do. I ask what I need to understand the problem before suggesting anything.
Before I start, you receive in writing what I will deliver, what is included and what is not.
While I work, I share progress you can open and check, so you can see that it is going in the right direction.
You receive the code and the documentation agreed in the proposal.
If someone in your business copies data from Excel or PDF files into another system, checks it by hand and mistakes still get through, I can automate that work. I build tools that check the information, keep the process in order and let you review the data before it is saved.
I also build professional websites, so your customers understand what you offer and how to reach you.
I'm Luis Pimentel, a junior Python developer working remotely from Potosí, Bolivia, with Python, FastAPI and PostgreSQL, and with frontend tools such as Next.js when a project needs an interface. I work in Spanish; in English, our communication will be mainly written.
Only what these projects actually use, and where you can see each tool.
We agree on both in the written proposal, before I start. The websites I build are static files, so they can be published on many hosts, including free static hosting such as GitHub Pages.
For a website: your texts, your logo, the photos you want to use and your business details, such as what you offer, your hours, where you are and how to contact you. For a data tool: example files with the real data removed or replaced, and a short description of what you do with them today. Please don't send passwords or sensitive personal data.
I reply within 24 hours. I work in Spanish; in English, our communication will be mainly written.
Tell me what you need. The form prepares an email for you; nothing is sent from this page.
Available for freelance projects, remote.
Please don't include passwords, credentials or sensitive personal data in your first message.
Found me on Upwork? Please keep all communication in Upwork Messages until a contract is in place, as Upwork's terms require.