I'm Biplov Bhandari -- I build cloud-native data and AI systems across public and private sectors, often for environmental and infrastructure programs.
Google Developer Expert in Google Cloud (Cloud AI) · Google Cloud Certified Professional ML Engineer · Currently a Senior Data Engineer on NOAA's Next Generation Water Prediction Capability (NGWPC) program.
My work runs the full path from raw data to decisions: designing warehouses and pipelines, orchestrating distributed compute, training and deploying models, and getting the results in front of people who act on them.
- 🏗️ Data Platforms & Pipelines -- BigQuery warehouse design, ETL/ELT, batch and streaming, orchestration, data quality and lineage
- 🤖 ML & AI Systems -- model training to production deployment, foundation models, embeddings and vector search, LLM/VLM, agentic workflows
- 🛰️ Geospatial at Scale -- petabyte-scale Earth observation, LiDAR, SAR, cloud-native formats
- 🗣️ Talks, Workshops & Open Source -- conference sessions, technical writing, tutorials, open-source tooling
Domains: environmental hazard and physical risk · water and flood prediction · climate resilience · agriculture · community-based participatory mapping
| Project | What it is |
|---|---|
| weathernext-outage-forecasting | BigQuery solution joining Google DeepMind WeatherNext forecasts with DOE EAGLE-I outage data to score county-level utility outage risk. BQML models, partitioned/clustered tables, query-cost controls. |
| earthrise-book-assistant | Quarto book viewer and FastAPI RAG assistant with hybrid retrieval, citations, tests, and containerized deployment. |
| g4g-25-demos | BigQuery vector search over AlphaEarth embeddings. GeoParquet from S3 to GCS to BigQuery, IVF indexes, semantic search from an Earth Engine front end. |
| NGWPC/bridge-classification | Automated LiDAR bridge-deck classification using USGS 3DEP data. Weak-label generation with PDAL/SMRF/RANSAC, sparse 3D U-Net, GPU inference on AWS Batch across ~220K bridges. |
| servir-aces | Open-source Python package for agricultural classification from satellite imagery (DNN/CNN/U-Net). Published in JOSS. |
| stac-lite | Lightweight, zero-infrastructure STAC API and browser for exploring static STAC catalogs with CQL2 search and filter support. No database required. |
- Embedding Vector Search and Beyond with BigQuery, Earth Engine, and AlphaEarth Foundations -- published on Google's official Google Earth and Earth Engine channel
- Geo for Good 2025 (Google Earth Outreach, NYC) -- session on BigQuery vector search
- Co-editor, NASA EarthRISE Applied AI and Deep Learning Book
- TheGeoICT -- tutorials on cloud, BigQuery, and applied geospatial ML
- Google Scholar -- 18+ peer-reviewed publications
- Languages: Python, SQL, JavaScript, Java
- Data Engineering & Storage: BigQuery, PostgreSQL/PostGIS, MySQL, ETL/ELT, batch & streaming pipelines, data-lake & warehouse architecture, data governance & lineage, Parquet/GeoParquet, Zarr, COG, STAC
- Orchestration & Infra: Dagster, Prefect, Airflow, Cloud Composer, AWS Batch, Docker, Kubernetes, Terraform, CI/CD (Cloud Build, GitHub Actions)
- ML & AI: TensorFlow, PyTorch, scikit-learn, Keras, BQML, Vertex AI, deep learning, MLOps, RAG, agentic workflows
- Google Cloud: BigQuery, Dataflow, Pub/Sub, Cloud Composer, Vertex AI, Cloud Run, GKE, Earth Engine
- Geospatial & GIS: GDAL, PDAL, Rasterio, GeoPandas, xarray, Google Earth Engine, remote sensing, QGIS, ArcGIS
- Other: Technical writing, teaching, stakeholder communication, team leadership
LinkedIn · X · YouTube · Google Scholar · ORCID
Open to collaboration on geospatial, data engineering, and applied ML projects -- feel free to reach out.