Founder

Theo Bazille • Peerlist

A hub where anyone can create, share, and use interactive documents — all centralized in one workspace. Davia is a workspace for creating interactive documents — pages that mix text, data, and AI-generated components. Davia pages are “living documents”, they follow a unique architecture combining editable content with interactive components. Each page is self-contained: it holds your content, your interactive components, and your data. Think of it as a document you can read, edit, and interact with.This is a new authoring paradigm: code is generated and compiled on the fly as you work. Describe what you need and the page can become anything—custom UI, logic, and data flows—without fixed blocks or rigid templates.

Davia Workspace

Davia is the fastest way to build internal tools tailored to your team’s unique workflows. It connects to your existing data sources—like HubSpot, Zendesk, Google Sheets, and more—and lets you design custom apps just by describing what you need. Whether it's a dashboard to triage support tickets, a sales tracker, or a reporting tool, Davia turns your prompt into a working interface in minutes.With built-in AI capabilities, Davia can automate repetitive tasks like tagging, prioritizing, or summarizing data. Your tools stay flexible and up to date as your processes evolve—no need for months-long development cycles. It’s software that adapts to your team, not the other way around.

Davia

- Developed an algorithm implemented on risky infrastructures (suspension bridges) to guarantee users’ safety.
- Used various processing techniques (Deep Learning, SVM, Random Forest).

Detection of cable failures by processing and classifying ac

- Research project on instabilities and non-linear phenomena in hydrodynamics. 
- Studied the behavior of a fluid (water) in the presence of vortices. 
- Enhanced the accuracy in modeling results.

Experimental Project in Fluid Dynamics

- Evaluated the effectiveness of classical ML models from Ridge Regression to Gradient Boosting (XGBoost)
- Built different architectures of RNNs and backtested them on more than 20 companies

Forecasted company’s quarterly earning using alternative dat

- Improved the efficiency of Monte Carlo methods on the calculation of an expectation of the function of Gaussians by quasi-Monte Carlo methods.
- Simulated uniform laws in a 5 dimension hypercube by Halton sequencing.
- Generated independent Gaussians using the Box-Muller method.

Optimisation of the calculation of an expectation by quasi-M

- Calibration and sensitivity analysis of model parameters on French regional data.
- Determined the Sobol indices by Monte Carlo method.

Quantified analysis of the SEAIR model applied to the covid-

- Synthesized metal salen complexes under strict experimental conditions (oxygen deprived atmosphere).
- Highlighted the mechanisms that govern the transport by using a simpler molecule:  Co(salen) complex.