Freelance projects

Freelance projects

General geospatial database

Databases & SQL, Python — incorrectly specified categories?

Project translated automatically. Log in or register, to view the original

Create a geospatial database that combines several public sources (DVF, ADEME, Cadastre, BAN, etc.) for the identification and matching of real estate objects even with partial or approximate data (coordinates, descriptions, or street photographs).

Scope of Work:

Data Collection and Normalization:

DVF (sales history)
Cadastre (geometry, cadastral parcels)
BAN (national address register)
ADEME (energy data / DPE)
Online listings (archive of 1–2 years)
Mapillary / Google Street (geotagged street and facade photos)

Task: build a unified repository linking parcels, buildings, and addresses.

Implement:

spatial joins and fuzzy matching of addresses/coordinates;
a clean and user-friendly API for internal use (FastAPI or GraphQL).

Final Use Cases for the Combined Database

Complete Listing → Exact Candidate for Building
Input: complete online listing with text (e.g., “2-room, 55 m², balcony, view of Sacré-Cœur, near Abbesses metro”) + photographs (including view from the window).
Process:
- extract approximate address/orientations from the description (NLP);
- match GPS from photo metadata (if available);
- cross-check view photos with the geotagged photo database (Mapillary, Google Street View);
- narrow down the list of buildings by cadastral polygons.
- enrich with ADEME data (energy, year of construction) and DVF (transaction history).
  Output: ranked list of candidates with values ban_id, id_parcelle, address, and trust score.
Approximate Coordinates + Listing Text → Best Matches
Input: listing with GPS, but with an error of ±300 m.
Process:
- find all parcels/buildings within the buffer zone;
- filter by listing characteristics (floor, area, number of rooms, presence of a balcony, year of construction) using DVF + ADEME + cadastral attributes.
  Output: shortlist of likely addresses with probabilities.
Cross-Search of Historical Listings
Input: current listing “loft 90 m², Marais, Paris”.
Process:
- matching with the archive of listings;
- checking if the object appeared earlier with an exact address;
- validating consistency (area, price, photo).
  Output: possible exact match with building ID and price history.
Recognition of View from Window
Input: photo from the window with a view of the Eiffel Tower.
Process:
- determining orientation and recognizable landmarks in the image;
- cross-checking with geospatial data (which buildings have such a line of sight);
- considering floor/height from the cadastre/INSEE.
  Output: list of building candidates corresponding to the perspective.
Detection of Suspicious/Fraudulent Listings
Input: listing claims “30 m², €70,000, near Gare de Lyon”.
Process:
- cross-checking with DVF for buildings in that area;
- comparing claimed area with cadastral records;
- detection of price anomalies.
  Output: flag “Unrealistic price, possible fraud”.
Identification Without Address
Input: listing only with description: “Studio 18 m², near Place d’Italie, quiet street, 5th floor, built in the 1970s”.
Process:
- NLP extracts the district and approximate year of construction;
- filtering BAN + ADEME for buildings in the 13th arrondissement, built around 1970;
- applying area filter based on DVF.
  Output: set of 2–3 candidates with trust scores.
Multi-Source Enrichment
Input: address or discovered building ID.
Process: link all databases (DVF, ADEME DPE, cadastre, normalized BAN, historical listings).
Output: a single object card with consolidated attributes (transaction history, energy indicators, building geometry, geotagged photos).

✅ Expectations from Data Engineer:

Build a normalized data warehouse (BAN + DVF + ADEME + cadastre + photographs).
Implement matching pipelines (spatial join, fuzzy text matching, photo verification by geo).
Provide an API that accepts raw inputs (listing text, coordinates, photos) and returns ranked candidates for addresses/buildings.

Proposals 3

Please select

Maksym O.

5 0

Budget: 1000 EUR Deadline: 1 day

Hello, I worked on a **geospatial database** - processed 5 sources, 3000+ objects, integrated an API for location-based query responses.

How do you plan to implement fuzzy address matching to improve accuracy?

I suggest we get in touch, I will provide you with free technical consultation and we can create a development plan + I will tell you about my team!

Stanislav K.

6 0

Budget: 3500 EUR Deadline: 60 days

Hello! 👋

I have solid experience building geospatial data systems, ETL pipelines, and fuzzy-matching algorithms for real-estate and urban analytics.
Your project is fascinating — merging DVF, ADEME, Cadastre, BAN, and photo sources into a unified geospatial knowledge base for property identification.

📌 Here’s what I can deliver:
• Build a normalized PostGIS data warehouse, harmonizing keys between DVF, BAN, ADEME, and cadastral data.
• Implement matching pipelines: spatial joins, fuzzy address matching, EXIF geo-extraction, and NLP text parsing (spaCy / transformers).
• Integrate data enrichment from ADEME, DVF, Mapillary/StreetView, and historical listings.
• Provide a clean FastAPI (or GraphQL) endpoint that accepts text, coordinates, or photos and returns ranked property candidates (ban_id, id_parcelle, confidence score).
• Add a data-quality layer: logs, anomaly detection (price, area, duplicates), and matching statistics.

🧠 Stack: Python (pandas, GeoPandas, shapely, scikit-learn, OpenCV), PostgreSQL + PostGIS, FastAPI, Airflow, Docker.
⏱ Timeline: 6–8 weeks for MVP (data warehouse + API).
💰 Budget: from $3,500 — depending on data scope and required automation depth.

I’d be glad to discuss the data architecture and show a concept demo of the BAN + DVF + Cadastre integration.

Gayane M.

0 0

Projects -
Rating -
Rating 162

Budget: 25 EUR Deadline: 1 day

Hello.
I am interested in your project. I am confident that I can deliver effective and high-quality work that meets your requirements and expectations. I have over 8 years of experience. I am ready to discuss the details and start working. I look forward to your response.

Tony Aliyev
France
Paris Management Group

Projects 111
Rating 5.0
Rating 8 531

General geospatial database

Scope of Work:

Final Use Cases for the Combined Database

Maksym O.

Stanislav K.

Gayane M.

Proposals are currently absent

Current freelance projects in the category Databases & SQL

Order form 1C

Microsoft 365 / Modern Workplace Consultant (m/w/d) – Microsoft Places | Remote

Admin panel extension: adding SEO service pages

Development of a CRM for processing orders from Prom, Rozetka, and own websites

Set up synchronization of two data sources with the CRM and eliminate desynchronization.