Molecular biologist · Developer · Product owner

Lucia De MojàFrom the lab bench to the backlog.

Lucia has designed lab experiments, written production software for six years and now works on AI products. The method never changed: decide what would prove you wrong, then go and look for it.

6
years writing production software
11
person team, delivery built from scratch
3
Developer of original apps
3
languages: Italian, English, Swedish

Featured

Pinned work

The pythonforbeginners.org home page: Master Python, One Line at a Time

luciademoja-lab/python-for-beginners

Python for Beginners

The companion learning platform to the book Python Made Simple: 97 interactive exercises that run in the browser, progress across six phases, and a free 1:1 coaching session for every reader.

97
interactive exercises
6
learning phases
~140
automated tests
Read the case study
The Piano Agricolo public mushroom catalogue, marked as provisional until health authorisations arrive

luciademoja-lab/piano-agricolo

Piano Agricolo

One dashboard for a planned farm in Cinquefrondi, Calabria: crops, planting windows, cash flow in three scenarios, a mushroom pricing model and the regulations behind every number.

3
cash flow scenarios
1 month to 3 years
projection range
2
access roles
Open the catalogue
The Elayra home page: You do not need to know what to write, start from how you feel

elayra.app

Elayra: The Cure

A spiritually inspired web experience for self-love, healing and collective transformation: a journal, daily light, plans and a community space, starting from how you feel in under a minute.

7
spaces to explore
1 min
to get started
TS
TypeScript
Visit the site
luciademoja-lab/elayra-researchOngoing

Attention weights and quantization

Transformer research · 2026

Research into how attention layer weight distributions differ across transformer models, and what that implies for quantizing them.

  • Python
Research
View on GitHub
TATAA Biocenter · GothenburgThesis

Artifacts in qPCR

Master's thesis, molecular biology · 2014

Investigating unspecific products in quantitative PCR: the false signals that look like results until you design the controls that expose them.

  • Lab protocol
  • Research
Research
Swedish University of Agricultural Sciences · SkaraLab

Fungal DNA from soil

Laboratory work · 2015

Preparing soil samples for the extraction and purification of fungal DNA contained in spores.

  • Lab protocol
Lab

All work

The path

Four stages, one method.

  1. 01

    Bench

    2007 to 2015

    Animal sciences, veterinary medicine and a master's in molecular and infection biology. qPCR, DNA extraction, experiment design.

  2. 02

    Code

    2016 to 2022

    Six years of production software in Gothenburg: Ruby on Rails, GraphQL, React, health data at Glooko.

  3. 03

    Product

    2022 to 2024

    Sole product owner for a blockchain launchpad, delivery practice built from nothing for an 11-person team.

  4. 04

    AI

    2025 to now

    Transformer and quantization research, AI governance, and edtech products built end to end.

Plate map

Skills, laid out like an experiment.

Ninety-six wells, one skill per filled well, coloured by channel. Tap or point at a well to read it, or filter by channel.

Product ownershipProduct channel

  • Product ownership
  • Backlog and user stories
  • Scrum and Agile
  • Capacity planning
  • Roadmaps and SRS
  • Stakeholder demos
  • AI governance
  • Regulatory scoping
  • Python
  • TypeScript
  • React
  • FastAPI
  • PostgreSQL
  • Supabase
  • Ruby on Rails
  • GraphQL
  • Stripe
  • Testing
  • Transformers
  • Quantization
  • Code review
  • Experiment design
  • Bioinformatics
  • qPCR
  • DNA extraction
  • Infection biology
  • Molecular biology
  • Teaching Python

Method

How the work gets done.

01

Ask what would prove you wrong.

Designing an experiment and specifying a product are the same discipline: decide in advance what result would mean the idea is wrong, then go and look for it.

02

Null is not zero.

A missing number is an honest gap. A made-up zero is a quiet lie that spreads through every total built on it. Label estimates, keep sources, show what is missing.

03

Talk to engineers without a translator.

Six years writing production software means requirements are written with the implementation in mind, and trade-offs get discussed in the team's own language.

04

Look after the people doing the work.

A weekly optional call about workload and life, not tickets. Everyone came, every week, for two years. It was also the best early warning a planner could ask for.

Building an AI or technical product?

Lucia is looking for a Product Owner or Technical Product Manager role, remote in Europe, with a team that wants someone who can talk to engineers without a translator.

Say hello