Shubhanshu
Bishwash
AI & Geospatial Engineer — Generative AI · Climate Intelligence · Earth Observation
Architecting autonomous multi-agent workflows and deploying Vision Transformers to turn terabytes of raw satellite and climate data into decision-grade intelligence — currently running climate risk automation for MSCI's global issuer universe.
Agentic AI & LLM Toolchain
A.01Local LLMs, MCP servers, and multi-agent orchestration powering no-touch data collection and validation workflows that run end-to-end without human intervention.
ML & Simulation
A.02Time-series forecasting and ESG scoring at institutional scale, with Bayesian inference for uncertainty quantification and Monte Carlo stress testing across asset classes.
Geospatial Technology
A.03Earth observation, DEM/SAR fusion, and cloud-native geospatial pipelines processing terabytes of Sentinel, Landsat, MODIS, and VIIRS data for risk and urban analytics.
Climate Risk & ESG Analytics
A.04Automating physical risk and emissions scoring across the MSCI ACWI issuer universe, aligned to TCFD and EU Taxonomy disclosure frameworks.
Agentic Climate Risk Intelligence System
Multi-agent RAG pipeline ingesting 100k+ climate datasets into a vector store for natural language querying of physical risk scores, with Bayesian uncertainty quantification and Monte Carlo stress testing for derivatives desk positioning.
Satellite-Based Exploration Platform
Replaced drone/on-site surveys with open-source satellite data to track vegetation regrowth and oil slicks — Vision Transformer models at 92%+ accuracy across 400+ reclamation sites.
Automated GIS Pipeline — NYU Stern
Python/GDAL/PostGIS pipeline for NYU Stern's Project Mumbai, saving 1,200+ manual hrs/year, with GAN-enhanced legacy imagery and cloud-native flood risk simulation on GCP.