I’m Jade Meskill, a software engineer with 2+ years of experience building backend APIs and data pipelines for government-funded research in global supply chains and U.S. water resources; experienced with geospatial data science, graph databases, high-performance computing, and applied machine learning. Open to relocation.

Work Experience

Research Software Engineer @ Northern Arizona University

Jan 2025 - Nov 2026

USGS Water Accounts: technical lead on Developing the Next Generation of U.S. Water Accounts, a USGS 104(g) National Competitive Grant (2024-2026).

• Developed datasets and algorithms for 5 hydrology and machine learning manuscripts; 1 under review, 4 slated for 2026-2027.
  • Adapted a spatial disaggregation pipeline to produce 1 km CONUS data across 4 water use sectors, in the NAU HPC environment (Keras/TensorFlow, SLURM).
    • Cut compute time ~7.5× (5 h to ~40 min per raster layer) with vectorized patch reconstruction, per-zone sums, and prediction masking.
    • Diagnosed a mis-scaled loss function, bringing R^2 from -0.38 to 0.87.
• Co-planned project phases and coordinated 6 undergraduate researchers building CONUS water use datasets as inputs to System of Environmental-Economic Accounting (SEEA) water accounts.

FEWSION Supply Chain Database

• Designed a provenance-based consensus system for merging conflicting facility records from separate data providers, with per-field lineage, confidence, and trust-weighted consensus functions (PyNest, MongoDB/beanie, Azure CosmosDB).
  • Synced consensus records from MongoDB into PostgreSQL for spatial queries.
• Developed an entity-resolution matcher to link incoming records to existing facilities, avoiding duplicates and scoring candidates on industry code, facility type, fuzzy name similarity, and distance (H3, async SQLAlchemy).
• Built Azure Functions ingestion pipelines for millions of records across economic sectors worldwide: Premise field surveys, Overture Maps (DuckDB), OpenStreetMap infrastructure, and global ports of entry (land, sea, and air).
• Migrated my generic CRUD API layer to high-availability PyNest services and async SQLAlchemy.
• Authored an internal library of parameterized Cypher queries in Neo4j to feed an economic network visualization with multi-hop trade flow metrics and trade relationship resilience.

Research Intern @ Northern Arizona University

Jan 2024 - Dec 2024

USGS Water Accounts

• Modeled environmental flow requirements for ecosystems across the Potomac River Basin for a high-resolution water tower model, through kriging interpolation of USGS stream gauge records (PyKrige).
• Developed a mass-preserving spatial disaggregation pipeline for monthly CONUS thermoelectric power plant water use, allocating each plant’s consumption and withdrawal by building footprint area (GeoPandas, Shapely, rasterio, Overpass API).

FEWSION Supply Chain Database

• Built a MongoDB REST API using beanie and an accompanying pipeline to preprocess Overture Maps data into provenance-tracked data sources.
• Designed a generic CRUD and routing layer for a FastAPI + SQLAlchemy/GeoAlchemy2 REST API over PostGIS, with spatial and AND/OR filtering and pytest.
• Implemented a global boundary reconciliation algorithm (GeoPandas, Shapely) that detects slivers where national and subnational boundaries fail to nest and merges them into neighbors.

Student Support Technician @ Northern Arizona University

Aug 2021 - Dec 2024

• Resolved over 4,500 call and email tickets campus-wide across identity management, email, and software licensing.

IT Specialist @ D2 Surveying

Oct 2020 - Oct 2021

• Bootstrapped a FreeBSD FTP server for field file transfer.
• Deployed a peer-to-peer Syncthing network for office collaboration.
• Wrote Python and VBA Excel automations to organize customer and geospatial data.

Projects

genixis

A Nix flake for declarative homelab and development infrastructure across macOS, WSL, and NixOS machines.

• Reproducible hosts: one set of modules builds every machine; per-machine settings stay out of the repo, and disko handles partitioning. Configured machines can be rebuilt with a single command.
• Self-hosted services: Podman containers for a Git forge (Forgejo), a Matrix homeserver, a password manager, and a media stack; exposed through Caddy and Tailscale, with agenix-encrypted repo secrets.
• CI: a nightly Forgejo Actions job on a self-hosted runner updates all inputs, checks formatting, then evaluates and builds every configuration in the Nix sandbox.
• Agent workflow: per-branch git worktrees for running parallel agents in one repository. Context, skills, and memory are version-controlled and generated from the same flake.

Education

Bachelor of Science in Applied Computer Science @ Northern Arizona University (Flagstaff, Arizona), Dec 2024, GPA 3.67

Publications

• Sjöstedt, E. C., Meskill, J. I., Kearns, M., Rushforth, R. R., & Zegre, N. P. (2026). Basin-Scale Water Towers: Insights from Comparing High-Resolution Water Tower Models of the Potomac River Basin, USA. Under review; preprint on ESS Open Archive. doi:10.22541/essoar.15006823/v1

Skills

• Languages: Python, SQL, Cypher, Nix, Bash, VBA
• Backend & APIs: FastAPI, REST, Pydantic, SQLAlchemy (async), GeoAlchemy2, beanie, PyNest, psycopg2, pytest
• Databases: PostgreSQL/PostGIS, MongoDB, Azure CosmosDB, Neo4j, DuckDB
• Data & ML: ETL pipelines, entity resolution, fuzzy matching, pandas, NumPy, scikit-learn, Keras, TensorFlow, Matplotlib, Jupyter
• Geospatial: GeoPandas, Shapely, rasterio, H3, OpenStreetMap/Overpass API, spatial disaggregation, ordinary kriging (PyKrige)
• Infrastructure: Linux, FreeBSD, Git, Nix Flakes, NixOS, nix-darwin, Home Manager, disko, agenix, Docker/Docker Compose, Podman, Proxmox, Caddy, Tailscale, CI/CD (Forgejo Actions), Azure (Functions, Blob Storage), HPC, SLURM, GPU computing
• AI tools: Claude Code, parallel agentic development workflows