Sarah Le Net

AI for resilience projects: development of a platform to prioritize and plan for projects and reconstruction after natural catastrophes, and IoT data acquisition, project-based – Fine-tuned Mistral reasoning and Claude report generation LLM pipeline with Qdrant vectorstore, climate and remote-sensing data and Langchain agents deployed on GCP Cloud Run – Interviewed on a blog around 25 people in emergency crises, NGOs, and administrations – Participated in 100+ networking and tech watch meetups and conferences – Started courses on remote sensing, emergency crises and risk management at Panthéon-Sorbonne, and metrology and citizen science at Conservatoire National des Arts et Métiers – Won a prize at Women In Data Science for an AI irrigation project, 2nd prize at HackTheGap for ResilienceAI, participated in 15 hackathons, found an incubator, and potential partners – Presentation at Hors Normes (ENS), Drive TLV and X-Up (bac à sable)

– GeoHack AI: locust breeding ground prediction with Sentinel imagery on Kaggle - selected to present solution with InstaDeep algorithm – Doctolib: Proposing a solution of LLM + RAG (adding articles and biomedical data) to produce a diagnostic and appointments to specific specialties – WIDS (Women in Data Science): On ethics and explainability Early maker prize from EMLYON business school: Smart irrigation system with LLM + RAG + RL and GAN predictions to tackle food security and sustainability in water management and fair ressources repartition – Microsoft AI agents: Emergency Guardian Agent (button trigger on your phone or with whom you could chat). The Agent would then act upon your emergency or advice: call emergencies, call SOS médecin, call your caretaker, take an appointment, call the police, create a transcript of your voice input – IBM watson X: Social media for emergency crisis agent, Relief to Resilience, which is aimed at proposing focused projects after a disaster to cope with the most important, impactful parameters; cartography, RAG – FrugalAI with HuggingFace: deforestation sound classification on the HuggingFace platform (API creation) with codecarbon to reduce the AI footprint – Bat'Adapt: OpenBooster competition, evaluation of LLMs for documentation on building safety in climate change – ESPF@Ekimetrics: Predictive Maintenance of railways, with survival models attempts – Google Hackathon in July: Healthcare solution – IBM hackathon in August: Social Media, WhatsApp servers for emergency crisis – Les clés du Château de Versailles (Agents): datacraft. Arrivés 4ème – BlackBox AI @42: one project on real-estate research. Another on TheLab – Pioneers AI @ Station F: Resilience AI – HackTheGap (AI for Good): 2nd Prize @ 42, Pioneers AI for a Resilience AI – Iterate x Columbia AI Club: Brainwave, a brainstorming voice assistant – June: Acted in Terrasses de Laurent Gaudet at the Jeux Isthmiques and Espace Beaujon

Equivalent to Master's degree in MLE with Sorbonne University Teacher with masterclasses and Q&A sessions for master levels in partnership with La Sorbonne: – Data Sciences (Interpretability, Text Mining, Deep Learning, Voice Recognition, Multibox Detection, Computer Vision, production tools) – Data Engineering (Intro and Streaming of Data) – MLOPS (Life cycle management of a machine learning model) – Data Analysis in bootcamp or continuing training in B2B or B2C. Other responsibilities and achievements: – Creation, translation and improvement of educational content (Couchbase, TechAway with OMNES (Data Quality, Machine Learning Ivl 3) – Cohort leader (4 to 5 cohorts permanently in 3 months or 1 year span) – Management of learners at the technical and informational level – Project group mentoring (15 projects permanently) (more than 80 projects overall) Research and exploration of new technologies for content creation – Participation in the educational reorganization of courses (creation of experimental and personal research course formats) – Internal analytics (on pedagogical data) – Creation of 2-hour workshops to feed the Youtube channel feed (PCA, multibox detection, multiprocessing, deployment of a ML model from scratch) – Writing articles on technologies associated with data (Altair, PyTorch Lightning, Theano) – Exam corrections – Google Cloud Skills Boost Certification - DE and Cloud Architect – Participation in internal training on various technologies (Make, Docker, Spark, etc.) – Integrating in a team workflow (team of 20 people in the teaching pole)

A coworking space with daycare in Paris, to facilitate access to work for single parents - Early-stage project, looking for investments, prospective for firm, part of the collective Empow'her for women's entrepreneurship – Earned a prize at GEF (Grandes Ecoles au Féminin) for the idea

concrete start of journey towards TheLab – Acquaintance with Datakind and other pro-bono projects – Kaggle competitions – Meetings with other volunteers – Giving advice and corrections to Data Engineering workbook – Networked to know the actors of the social field – Inquired about GenAI and prediction tools in Humanitarian crises be it on the climatic or logistic side

Participation in : – Bloom project for boat traffic and fishing in protected areas (addition to database and research on characterizing fishing patterns in boats and discussions on fishing patterns recognition) – Pyronear: benchmarked raspberry pie solutions and adaptation, designed a blurring algorithm for people on the forest pictures – Potentiel Solaire: explored the SAM tool

Travail sur un application streamlit pour le Water-Tracker de LaReserveTech https://github.com/LaReserveTech/water-tracker/blob/main/streamlit.py

As auto-entrepreneur Fabrica Sesame, Teach'r, Heyprof, Superprof: – Collective classes – Individual visio-conference – Homeschooling

– Generated baselines – Reported and predicted of electrical consumption on air conditioning with a connected device – Designed reduction schemes and partnership with distribution companies – Start of research on predictions with SVM and Salesforce data Skills: Python, SQL, Apache Spark, C#.

– Learned coding – Researched on problematics around data for governance – Participated in social events for governance – Interviewed Directors of big companies such as Bloomberg head of Data on the Future of Data Science.

'Mining & Water Risk: Diagnosis, Benchmarking, and Quantitative Analysis of Financial Impacts' – Data analysis to build a Bayesian Inference to link water risks (pollution, flood) to portfolio investments – Data research

Center for International Earth Information Network (CIESIN) Decadal Geospatial Distribution of Historical Emissions SO2 Sub-National Level: Time Series From 1850 to 1970 Skills: Data Management with SQL and VBA on Microsoft Access, Mapping with ArcGIS

– Created a software to pull out scanning patterns with big data collected with ScanLife using SQL and VB.Net – Wrote a report on QR code markets