↗ 4 projects available
Find your next collab project
Connect with builders working on meaningful projects. Build your skills, and ship real things together.
Showing 4 projects
Most recent
berges.louis1
Multisensor of building - Temperature, humidity, CO2 levels measurements, electricity
Regulatory energy assessment methods (DPE, RTEx, RE2020...) estimate the theoretical energy performance of buildings under standardized operating conditions. While this approach enables objective comparison between buildings, it does not aim to predict actual energy consumption. The discrepancy between calculated and measured performance, commonly referred to as the "energy performance gap", arises from factors such as occupant behavior, control strategies, commissioning quality, maintenance practices and weather variations. The objective of the project is to leverage low-cost IoT sensors (M5Stack) to monitor actual building operating conditions and compare measured energy performance and comfort indicators with simulation-based predictions. By quantifying measurement and model uncertainties, the project aims to identify and prioritize cost-effective energy efficiency improvements, including operational adjustments and retrofit measures (with quantification of uncertainties). The following steps will be taken: 1) Real-time data acquisition and visualization Continuous monitoring of electricity consumption, indoor temperature, relative humidity, CO₂ concentration, outdoor temperature, outdoor humidity and solar radiation using M5Stack-based IoT sensors. Data will be collected, stored and visualized through a dedicated dashboard with plots including uncertainties. 2) Building energy modelling Development of a simplified thermal RC model and/or a detailed building energy model using dedicated simulation software (e.g. Pleiades). The model will be calibrated using available building characteristics and operational assumptions. 3) Comparison between simulated and measured performance Comparison of predicted and measured energy consumption, indoor temperatures and comfort indicators. The discrepancies observed will be used to quantify the building energy performance gap. 4) Model calibration and validation Adjustment of model parameters using measured data to improve the agreement between simulations and real operating conditions. The calibrated model will provide a more representative description of the building's actual behaviour. 5) Uncertainty quantification Assessment of uncertainties associated with sensor measurements, weather conditions, occupant behaviour and model assumptions. Uncertainty propagation techniques will be used to evaluate the confidence intervals of model predictions and performance indicators. 6) Sensitivity analysis Identification of the parameters having the greatest influence on energy consumption and thermal comfort (e.g. insulation level, air infiltration rate, heating setpoint, ventilation strategy). 7) Identification of improvement measures Evaluation of potential energy efficiency measures, including operational improvements (setpoint optimization, ventilation scheduling, occupant awareness) and retrofit solutions (insulation, glazing replacement, HVAC upgrades). 8) Cost-benefit analysis and decision support Estimation of energy savings, comfort improvements, implementation costs, payback periods and associated uncertainties for each proposed measure. Recommendations will be ranked according to their expected cost-effectiveness and robustness. 9) Development of a decision-support framework Integration of monitoring, simulation, uncertainty analysis and economic assessment into a reproducible methodology applicable to other residential or tertiary buildings equipped with low-cost IoT sensors
Eugenio La Cava
France
Circuit-JEPA
Goal — Build Schematic-Aware JEPA, a graph-based AI model that learns electronic-circuit structure through self-supervised pretraining and acts as a semantic co-pilot for an EDA tool, performing functional-block recognition and semantic design-rule checking. Components — Five layers: data sourcing (atopile registry, KiCad/Altium corpus, synthetic-negative generation), ingestion and canonicalization, graph representation, JEPA pretraining (context/target encoders, predictor), and EDA integration with equivalence-aware evaluation. Team — Roughly 6–8 people: a research lead, two graph-ML / self-supervised researchers, one or two data engineers, an electronics/EDA domain expert, an MLOps engineer, and a software engineer for tool integration. Skills required span graph neural networks, JEPA, Python data pipelines, circuit design and SPICE, and large-scale training infrastructure. Duration — Roughly 18–24 months to a validated prototype: about 6 months on data and the oracle, 9 on representation and pretraining, the remainder on downstream tasks, integration and evaluation.
Eugenio La Cava
France
CCR-learning-router
A self-improving custom router for claude-code-router. Routes requests across three model tiers (S/M/L) using progressively more sophisticated decision policies. PoC at https://github.com/eugenio/ccr-learning-router
Richie Mouhouadi
France
Docka
DocKA is an open-source **Domain Knowledge Acquisition & Retrieval Platform**. It transforms raw, heterogeneous documents — PDFs, Word files, HTML pages, plain text — into a **clean, searchable knowledge base**, using a modular pipeline that is: - **Domain-agnostic** — works on any corpus: technical docs, healthcare literature, legal texts - **Idempotent** — safe to re-run without creating duplicates - **Extensible** — designed to grow from keyword search to semantic search to RAG DocKA is not just a search engine. It is a **knowledge acquisition system** — the infrastructure that makes documents *useful*.