Kalai Ramea

Resume

Kalai Ramea
← Back

Applied AI · Builder · Tinkerer

Senior ML researcher and engineer with 10+ years building and shipping production AI systems — from physics-informed transformers to agentic AI platforms. When not doing science, drawing comics of everyday things.

20+
Years experience
$4M+
Funding secured
30+
Publications
1,170+
Citations
Career timeline
Now
2026 — present
Independent Researcher
2026 — present
South Park Commons Fellowship · San Francisco, CA
  • AI Infrastructure Research: Investigating how token-level inference demand maps to GPU utilization, memory pressure, and data center capacity constraints.
  • Carnot Router: Designed and built an end-to-end inference routing system — from token generation through GPU and memory health, grid health, and environmental conditions — to intelligently route inference jobs to optimal data center locations based on real-time resource and sustainability signals.
  • Agentic AI Platform (Matter42): Designing and building an agentic AI platform for scientific simulations and defect analysis; leading end-to-end architecture from system design through implementation.
AI infrastructure Inference routing Agentic AI Ethiopia fieldwork
AI Industry
2017 — 2025
Co-founder & Principal ML Researcher (CTO)
2023 — 2025
Planette AI · San Francisco, CA
  • Research & Model Development: Designed and implemented physics-informed transformer and diffusion architectures for long-range weather forecasting in PyTorch, achieving 15% improvement over NOAA and ECMWF benchmarks; trained 100M+ parameter models on multi-GPU clusters; applied SHAP and attention visualization to interpret model predictions.
  • Production Infrastructure: Built end-to-end systems on AWS and GCP — 10TB+ data pipelines, REST APIs with <5s latency, containerized deployment (Docker, Lambda/DynamoDB/S3), ReactJS web and mobile applications.
  • Research Leadership & Funding: PI on NSF SBIR ($300K) grant; led cross-functional team of 5 ML engineers and climate scientists; co-founded company and raised $2.4M in venture capital.
VC $2.4M NSF $300K PyTorch Transformers AWS · GCP 10TB+ pipelines
Senior Manager & Lead Scientist, ML and Data Science
2020 — 2023
Palo Alto Research Center (PARC) · Palo Alto, CA
  • DARPA-Funded Research Program: Led $1M DARPA program applying AI to climate science; developed AiBEDO climate emulator (published CIKM 2023); deployed on HPC enabling intervention scenarios 1000× faster than traditional methods.
  • Production Systems: Built and shipped conversational AI chatbot for COVID-19 assistance (PARC Most Innovative Application Award 2020) and neural style transfer pipeline for commercial deployment (TensorFlow, Docker, AWS).
  • Research Program Management: Secured $500K NASA funding; directed team of 6–10; managed hiring and mentorship; published at ICML, AAAI, and IUI.
DARPA $1M NASA $500K CIKM 2023 HPC Team of 10
Research Scientist
2017 — 2020
Palo Alto Research Center (PARC) · Palo Alto, CA
  • Applied ML Systems: Developed computer vision system for AR repair workflows (YOLO-based, real-time mobile inference); conducted user-centered evaluation studies for conversational AI interfaces improving accessibility for visually impaired users.
  • Research Contributions: Published at ICML workshop, AAAI, and IUI; cross-lab collaborations at the intersection of ML, HCI, and climate informatics.
Computer vision YOLO Conversational AI ICML · AAAI · IUI
Academia
2010 — 2017
Postdoctoral Researcher
2016 — 2017
University of California, Davis
  • Developed ML and optimization frameworks integrating consumer behavior into large-scale energy systems models — a nonlinear, multi-agent system with long planning horizons; published in Nature Energy.
Nature Energy Optimization Energy systems
PhD Researcher — Transportation Technology & Policy
2010 — 2016
UC Davis · ITS & IIASA (Laxenburg, Austria)
  • Pioneered methods integrating behavioral elements in long-term energy systems models. Analyzed policy scenarios for California Air Resources Board. Selected as IIASA Young Scientists Summer Program Fellow 2013. International collaboration with University College Cork, Ireland.
IIASA Fellow 2013 Econometrics Climate policy CARB
Engineering
2005 — 2010
Engineering Consultant
2005 — 2010
Kimley-Horn and Associates · Oakland, CA & Iteris, Inc. · Anaheim, CA
  • Transportation planning and civil engineering consulting for public-sector infrastructure; data-driven decision-making through modeling, simulation, and spatial analysis.
ITS GIS AutoCAD Traffic modeling
Technical skills & expertise
ML & deep learning10+ yrs
PyTorch · transformers · diffusion models · physics-informed NNs · multi-GPU (A100, V100)
Production ML & cloud7+ yrs
AWS · GCP · Docker · REST APIs · CI/CD · W&B · MLflow · Zarr · Dask
Programming & scientific computing10+ yrs
Python · NumPy · Pandas · SciPy · scikit-learn · SQL · R · JavaScript/React
Data science & statistics8+ yrs
Bayesian methods · causal inference · time series · uncertainty quantification
AI infrastructure & systems3+ yrs
Inference routing · GPU/memory monitoring · data center capacity planning
Domain expertise10+ yrs
Climate science (CESM, ERA5) · geospatial/GIS · energy systems optimization
In the press
TEDx Gateway · Talk
How AI Could Help Forecast Floods Years in Advance
Watch on YouTube →
Tech Talks Daily · Podcast
Why Medium-Range Forecasts Could Save Millions
Listen on Apple Podcasts →
TWiML & AI Podcast
Time Series Clustering for Monitoring Fueling Infrastructure
Authority Magazine
Women Leading the AI Industry
Humans of ML · FloydHub
Exploring Computational Creativity with Neural Networks
Thrive Global
"Be creative." — Interview with Kalai Ramea
Highlighted publications
View all 30+ →
  • 2018
    Interaction of consumer preferences and climate policies in the global transition to low-carbon vehiclesNature Energy
    McCollum, Wilson, Ramea, et al.
  • 2023
    AiBEDO: Climate emulator for fast atmospheric intervention scenario analysisCIKM
    Ramea, et al. · ACM CIKM 2023
  • 2019
    Unsupervised Temporal Clustering to Monitor the Performance of Alternative Fueling InfrastructureICML
    Ramea · International Conference on Machine Learning
  • 2019
    Comparison of Transfer-Learning Approaches for Response Selection in Multi-Turn ConversationsAAAI
    Vig, Ramea
  • 2018
    Integration of behavioral effects from vehicle choice models into long-term energy systems optimization modelsJournal
    Ramea, Bunch, Yang, Yeh, Ogden · Energy Economics
  • 2015
    Achieving California's 80% greenhouse gas reduction target in 2050Journal
    Yang, Yeh, Zakerinia, Ramea, et al. · Energy Policy
30+ publications across Nature Energy, ICML, AAAI, CIKM, AGU, Energy Policy, Applied Energy, and more. Full list →
Honors & awards
  • 19
    US Frontiers of Engineering Symposium
    National Academy of Engineers — 1 of 100 young engineers nationally · 2019
  • 18
    Early Stage Innovation Excellence Award — AR Assistant
    Palo Alto Research Center · 2018
  • 17
    Early Stage Innovation Excellence Award — Conversational Agent
    Palo Alto Research Center · 2017
  • 15
    Helene M. Overly Memorial Graduate Scholarship
    WTS Sacramento · 2015
  • 14
    UCTC Dissertation Fellowship
    University of California Transportation Center · 2014
  • 13
    Young Scientists Summer Program
    International Institute for Applied Systems Analysis (IIASA) · 2013
  • 11
    ITS Davis-Chevron Fellowship
    Institute of Transportation Studies · 2011
  • 05
    Outstanding Academic Achievement Award
    University of Southern California · 2005