GISjobs.com – It All Starts Here

Robert Montalvo-West

Montreal, QC, Canada

Data scientist, historian, and disabled veteran working at the intersection of AI, geospatial intelligence, OSINT, and public-interest technology. Experienced building production-ready machine learning and remote sensing systems using satellite imagery, geospatial databases, and large-scale data pipelines. Combines interdisciplinary research training with hands-on technical implementation to produce accessible, policy-relevant analysis.

---

## Technical Skills

**Programming:** Python, SQL, Bash

**AI/ML:** PyTorch, computer vision and segmentation (YOLO, SAM, etc), LLM Integration, LLM-accelerated coding and agents, multimodal AI, machine translation

**Geospatial/Data:** PostgreSQL, PostGIS, GDAL, Pandas, Geospatial Analysis, Remote Sensing, Satellite Imagery Processing

**Infrastructure:** Docker, Docker Compose, Linux Administration (Arch, Debian, RHEL), Proxmox, Apache Kafka

**Tools:** Git, Hugo, Grafana, Cloudflare Tunnels, Cloudflare R2, n8n, Web Scraping/Crawling, Automation, Cron

**Research & Analysis:** Historical and policy research, critical thinking, complex pattern recognition, synthesis of large bodies of literature, legal and regulatory analysis, cross-disciplinary frameworks

**Communication:** Public speaking, persuasive writing, technical documentation for non-specialist audiences, policy brief production

---

## Relevant Projects

**[Sentinel Bird](https://sentinelbird.com)** (2025-ongoing) *Python, PostgreSQL, PostGIS, PyTorch, Remote Sensing*
- Designed, built, and deployed sentinelbird.com, a fully automated geospatial intelligence platform that continuously ingests, processes, and publishes satellite analysis of Gaza. Architected a zero‑touch pipeline that runs every 30 minutes, automatically discovering new Sentinel‑2 acquisitions, downloading the best available L2A product, and importing it into a PostgreSQL/PostGIS raster database. Integrated a PyTorch‑based semantic segmentation model to classify clouds and shadows at 10 m/pixel resolution, enabling accurate cloud‑free composites. Containerized the system with Docker, leveraging a PostgreSQL instance with a RAM‑backed tablespace for in‑memory raster processing and parallelized ingestion. Generated and distributed true‑color images and cloud/shadow overlays for every date, automatically creating compressed archives and uploading them to Cloudflare R2 for CDN delivery. Launched the public‑facing static site, serving georeferenced visualizations, timelapse videos, and complete downloadable archives to journalists, researchers, and content creators worldwide.

**[Synthetic Media Research System](https://rdmwest.com/projects/fifteen_gdelt/)** (2025) *Python, PostgreSQL, Kafka, Docker*
- Developed a fully automated research system that generates a synthetic news article every 15 minutes from real‑time multilingual GDELT event data, producing over 5,700 articles during a two‑month deployment. Orchestrated 12 Docker‑based micro‑services (ingestion, crawling, translation, LLM selection/summarization/generation, WebSocket updates) connected via Kafka and backed by PostgreSQL. LLM‑based filtering reduced hundreds of candidate source URLs per cycle to a curated set of ~15 for in‑depth summarization, with concurrent crawling and translation handling non‑English content. The zero‑touch pipeline operated for less than $1/day, demonstrating the viability of scalable, low‑cost automated disinformation and informing policy discussions on synthetic media, electoral integrity, and algorithmic amplification. Authored comprehensive documentation that translates complex multilingual event data into structured, accessible workflows.

**[U.S. Carceral System Mapping](https://rdmwest.com/projects/prison_industrial_complex/)** (2024) *Python, Google Earth Engine, PostgreSQL*
- Developed a systematic pipeline to map US prison facilities by distributing 50‑km‑radius search circles across the entire country and automatically querying the Google Places API for carceral‑related terms, producing thousands of candidate locations. Integrated an LLM‑based filtering step to retain only relevant establishments and retrieved high‑resolution NAIP satellite imagery for each confirmed site. Applied a multi‑modal LLM to classify each image (rural/urban) and extract structured metadata, building a labeled dataset suitable for training a custom YOLO object‑detection model. The automated workflow minimized manual curation and delivered coverage data positioned for public‑interest research, policy analysis, and future large‑scale detection of carceral facilities in satellite archives.

**Bridge Neural Network — Cross-Modal SAR/Optical Retrieval** (2025-ongoing) *Python, PyTorch, PostgreSQL, PostGIS, GDAL, Docker, CUDA*
- Built a self‑healing pipeline that ingests multi‑terabytes of Sentinel‑1 & 2 archives to train a cross‑modal retrieval model without manual steps, reducing a multi‑day GIS workflow to a continuously running system. A parallel extraction system with 55 concurrent GDAL workers and a 6 GB pre‑fetch buffer eliminates disk I/O during training, while a persistent written‑row mask allows interrupted extractions to resume without re‑work. The ~15 M‑parameter dual‑path ResNet encoder processes ≈14,000 triplets per epoch and is evaluated on 450 optical queries against a gallery of over 20,000 SAR tiles, achieving Recall@K and mean reciprocal rank across three temporal gap regimes (0‑6 d, 7‑30 d, 31‑120 d). A Thompson‑Sampling curriculum bandit automatically shifts difficulty from ~100 km spatial offsets to fine‑grained temporal matching, and a real‑time terminal dashboard monitors GPU/VRAM, loss curves, and auto‑saves logs every 30 seconds for remote inspection.

---

## Professional Experience

**Equity Director** — McGill History & Classics Graduate Students' Association (HCGSA) | 2019–2020
Held executive governance role within a McGill academic institution. Responsible for developing, interpreting, and enforcing institutional equity policy affecting a graduate research community. Navigated formal institutional processes and multi-stakeholder deliberation. Co-organized the 2020 McGill-Queen's Graduate History Conference.

**Infantryman, Specialist** — United States Army | 2010–2013
Team member and team leader in a high-pressure operational environment requiring exceptional attention to detail, clear communication, and disciplined execution under time pressure. Medically retired; disabled veteran.

---

## Education

**Master of Arts in History** — McGill University, Montréal, QC | GPA 3.80
- Graduate Option in Gender and Women's Studies
- MRP: *Death in the Archive* — Archival investigation synthesizing affect theory, institutional power, and historical methodology. Advisor: Dr. Brian Lewis, PhD

**Bachelor of Arts in History** — Montana State University, Bozeman, MT | Highest Honours, GPA 3.98
- Concentration in History of Science, Environment, Technology, & Society (SETS)
- Minor in Religious Studies
- Capstone: *Augustine: The Muted Manichean*

**Québec Corps de Francisation** — 44-week full-time French immersion program

**US Army Infantry School** — Basic infantry skills, fire team tactics, marksmanship, and small-unit leadership.

**US Army Airborne School** — Static-line parachute operations, aircraft exit procedures, airborne safety, and parachute landing falls.

**IBM Data Analysis Professional Certificate** — Data analysis, manipulation, and visualization, SQL, Python, and Jupyter Notebooks

---

## Public Speaking

- **Guest Lecturer**, McGill HCGSA — *The Borg: Star Trek and Assemblage Theory* (2019)
- **Guest Lecturer**, McGill HCGSA — *Augustine: The Muted Manichean* (2019)
- **Speaker** — Montana State University: 100th Armistice Day Anniversary (2018)
- **Panelist** — Montana State University Constitution Day: Campus Free Speech (2018)
- **Team Competitor** — 22nd Northwest Regional Ethics Bowl (2017)

---

## Awards & Recognition

**Academic:** Montana State University President's Honour Roll (2017–2018), Dean's Honour Roll (2017)
**Military:** US Army Certificate of Retirement (2013), Army Achievement Medal (2013), National Defence Medal (2011), Global War on Terrorism Service Medal (2011), Army Service Ribbon (2011)

Robert Montalvo-West

You must be a registered/logged in employer to contact job seekers.