Aadesh
Salecha

I'm a Founding Engineer at Hyfin, where I'm building GAIA, a platform that synthesizes asset performance and market data to power clean energy loan origination. I design the agentic data pipelines and risk intelligence tools aimed at closing the $4T+ clean energy financing gap across South Asia and Africa.

I recently finished my MS in Computer Science at Stanford, where I built geospatial ML for carbon sequestration with Prof. David Lobell's lab at the Doerr School of Sustainability, and studied human-like biases in LLMs (PNAS Nexus, WIRED). Before that, I worked on AI/ML at Amazon and large-scale financial systems at DraftKings.

I'm a Founding Engineer at Hyfin in San Francisco, where I'm building GAIA, a fintech platform that synthesizes asset performance and market data for clean energy loan origination. I design agentic data enrichment pipelines that combine LLMs and web scraping to create data for a new asset class with no track record, and I'm architecting risk intelligence tools aimed at the $4T+ clean energy financing gap across South Asia and Africa. I also lead the rollout of AI-native workflows across the team.

I completed my MS in Computer Science at Stanford University in 2026. There, I worked as an AI/ML researcher with Prof. David Lobell and the Center on Food Security and the Environment at the Stanford Doerr School of Sustainability, building geospatial ML pipelines on Google Earth Engine, Landsat, and Sentinel-2 imagery to detect tree species across cropland in the Global South. This work supports verification of the 5+ Gt CO₂/year sequestration potential of integrating trees into global cropland. I also won Stanford's first Green AI Hackathon by building a geothermal site prediction system on AlphaEarth satellite embeddings (93% AUC).

Earlier at Stanford, I worked with Prof. Johannes Eichstaedt on applying cognitive science and psychometrics to evaluate LLMs. Our paper showing that LLMs exhibit human-like social desirability biases was published in PNAS Nexus and covered by WIRED.

In 2025, I was an AI/ML Engineer at Amazon (AWS SageMaker HyperPod), where I designed an agentic AI system that combines telemetry, time-series signals, and RAG-based diagnosis to auto-resolve ML infrastructure failures, and cut GPU energy waste by about 200 kg CO₂e per day. Before Stanford, I was a Senior Software Engineer on the Financial Platform team at DraftKings, building production systems that handled billions of requests for millions of users. I received my BS in Computer Science, summa cum laude, from the University of Minnesota.