Steve Paul
AI Scientist  ·  Ph.D. Mechanical Engineering
New York, NY  ·  iamstevepaul@gmail.com
Steve Paul

About Me

I am an AI Scientist at Nantum AI, where I develop AI-driven solutions for building systems—including HVAC fault detection, air-side anomaly detection, LLM-based diagnostic recommendations, adaptive modeling for static pressure, Model Predictive Control, and Reinforcement Learning for air-side optimization—all deployed in AWS for real-world clients.

My work sits at the intersection of two core research pillars. The first is AI-based fault detection and prognostics across diverse physical systems: building HVAC infrastructure, industrial rotating machinery (bearings), and power distribution networks. I apply time-series analysis, Topological Data Analysis (TDA), Explainable AI (Grad-CAM), and survival analysis to detect anomalies, classify faults, and predict failures—enabling trustworthy, human-interpretable decision support in safety-critical environments.

The second pillar is intelligent planning, scheduling, and operations research for complex multi-agent systems. My Ph.D. (ADAMS Lab, University at Buffalo, advised by Dr. Souma Chowdhury) produced Graph Reinforcement Learning frameworks for Multi-Robot Task Allocation, Capacitated Vehicle Routing, Urban Air Mobility fleet scheduling, and power network reconfiguration—achieving real-time, scalable decisions under uncertainty that outperform or match state-of-the-art solvers at a fraction of the computation cost.

I also spent two years as a Software Engineer at Wayfair (Big Data, Apache Storm, AKKA, .Net). Outside of work, I enjoy reading, working out, hiking, piano, soccer, and astronomy.

Expertise

🔍
AI-Based Fault Detection & Prognostics

End-to-end AI pipelines for detecting, classifying, and predicting failures across physical systems—from building HVAC units to industrial rotating machinery.

HVAC Fault Detection Bearing Failure Prediction Anomaly Detection Survival Analysis Explainable AI Grad-CAM Time Series TDA LLM Diagnostics
🗺️
Planning, Scheduling & Operations Research

Scalable, real-time decision-making for complex combinatorial problems in robotics, logistics, and infrastructure—using Graph RL to match or beat classical solvers.

Multi-Robot Task Allocation Vehicle Routing (CVRP) Fleet Scheduling Network Reconfiguration Graph Neural Networks Reinforcement Learning Multi-Agent Systems MIP / LP
🤖
AI & Machine Learning

Deep expertise in modern ML architectures for sequential decision-making, state estimation, and trustworthy AI across safety-critical domains.

Deep Learning Transformers Policy Gradient RL Imitation Learning Probabilistic ML State Estimation Model Predictive Control PyTorch

Experience

Aug 2025
Present
AI Scientist
Nantum AI
  • Fault detection & diagnostics: AI-based HVAC fault detection and air-side anomaly detection from time-series sensor data; LLM-powered diagnostic recommendations deployed in AWS for client buildings
  • Planning & optimization: Adaptive modeling for static pressure set-point scheduling; Model Predictive Control and Reinforcement Learning for real-time air-side energy optimization
Feb 2025
Jul 2025
Research Scientist
Percev LLC
  • Led Percev's Analytics Team; survival analysis on bearings for failure prediction and defect classification
  • Architected CI/CD framework for analytics; Random Survival Forests with >75% accuracy
  • Commercialized prediction model as a new product; built customer-facing data visualization UI
Feb 2024
Jan 2025
Postdoctoral Research Associate
REIL Lab, University of Connecticut & Percev LLC
  • Fault classification & XAI: Explainable CNN (Grad-CAM) for bearing fault classification; lightweight trustworthy prediction framework with up to 22% accuracy gain for high-confidence samples
  • State estimation: Probabilistic high-rate dynamic system state estimation using TDA and ML; avg. compute time <50ms, outperforming state-of-the-art methods
  • Failure prediction: Survival analysis on rotating machinery; Random Survival Forests with >75% accuracy; commercialized as a product for Grace Technologies
Jan 2020
Jan 2024
Ph.D. Researcher
ADAMS Lab, University at Buffalo
  • Graph RL for MRTA, CVRP, UAM fleet scheduling, and power network reconfiguration
  • Funded by ONR AI/ML (N00014-21-1-2530) and NSF CAREER (CMMI 204802)
Jun 2018
Dec 2019
Software Engineer
Wayfair LLC
  • Notifications infrastructure: email & push for all Wayfair customers; personalized content engine
  • Supply chain routing (C# / .Net) for small and large parcel delivery across US and Canada
  • Built load-testing frameworks; maintained apps in AKKA, Apache Storm, Dropwizard, PHP

Education

Ph.D. in Mechanical Engineering
University at Buffalo, SUNY  ·  Jan 2020 – Jan 2024
Dissertation: Graph-based Higher-Order Graph Reinforcement Learning for Multi-Agent Systems and Physical Networks  |  Advisor: Dr. Souma Chowdhury  |  CGPA: 3.87/4
M.S. in Mechanical Engineering
University at Buffalo, SUNY  ·  Aug 2015 – Sep 2017
Dissertation: A Bio-inspired Neural System for Energy Optimal Collision Avoidance by UAVs  |  CGPA: 3.81/4
B.Tech in Mechanical Engineering
Mahatma Gandhi University, Kerala, India  ·  Jul 2010 – May 2014

Selected Publications

View full publication list on Google Scholar →

Nature Communications
*Jacob, R.A., *Paul, S., Chowdhury, S., Gel, Y.R., & Zhang, J.  (*joint first authors)
Nature Communications 15.1 (2024): 4766
ASME JCISE 2024
Paul, S. & Chowdhury, S.
ASME Journal of Computing and Information Science in Engineering (2024)
Robotics & Autonomous Systems
A Graph Capsule Attention Network to Learn Scalable Policies for Multi-Robot Task Allocation
Paul, S. & Chowdhury, S.
Robotics and Autonomous Systems (2023, Accepted)
ACM SAC 2024 – Best Paper Award
Graph Learning-based Fleet Scheduling for Urban Air Mobility under Operational Constraints, Varying Demand & Uncertainties
Paul, S., Witter, J., & Chowdhury, S.
ACM Symposium on Applied Computing 2024, AI & Agents theme
IEEE ICRA 2023
Efficient Planning of Multi-Robot Collective Transport using Graph Reinforcement Learning with Higher Order Topological Abstraction
Paul, S., Li, W., Smyth, B., Chen, Y., Gel, Y., & Chowdhury, S.
IEEE International Conference on Robotics and Automation (ICRA 2023), London, UK
IEEE ICRA 2022 – Outstanding Coordination Paper Nomination
Paul, S., Ghassemi, P., & Chowdhury, S.
IEEE ICRA 2022, Philadelphia, PA
Mechanical Systems and Signal Processing 2025
Probabilistic Machine Learning Pipeline using Topological Descriptors for Real-Time State Estimation of High-Rate Dynamic Systems
Chua, Y.K., et al.
Mechanical Systems and Signal Processing 227 (2025): 112319

Projects

Capacitated Vehicle Routing Problem (CVRP)
Graph RLCapsule NetworksCombinatorial Optimization

Scalable Graph RL framework for CVRP with up to 50% cost reduction and 20× faster computation vs. meta-heuristic methods, without retraining for larger instances.

Publication: ASME-IDETC 2022
Multi-Robot Task Allocation (MRTA)
GNNReinforcement LearningMulti-Agent

GNN + RL framework for MRTA with up to 6% higher task completion and 10× faster computation than meta-heuristics, scalable to larger problems without retraining.

Publication: IEEE ICRA 2022 (Outstanding Coordination Paper Nomination)
MRTA – Collective Transport
TDAGNNMulti-Agent

GNN policy augmented with Topological Data Analysis for multi-robot collective transport. Up to 10% higher task completion vs. other learning-based methods; decision time in milliseconds.

Publication: IEEE ICRA 2023
Urban Air Mobility Fleet Scheduling
Graph RLUAMFleet Planning

Learning-based UAM fleet scheduling with >12% revenue improvement over traditional RL, considering demand, passenger fare, and electricity pricing. Decision time in milliseconds.

Publications: AIAA Aviation 2022  ·  ACM SAC 2024 (Best Paper)
Power Network Reconfiguration
Graph RLPower SystemsOptimization

Graph RL for optimal network reconfiguration and power flow, achieving comparable performance to conventional methods with >10× faster computation.

UAM Vertiport Management
Graph RLUAMMulti-Agent

Graph RL for real-time vertiport take-off and landing scheduling under environmental uncertainties.

Publications: IEEE IROS 2023  ·  AIAA SciTech 2023

Awards & Honors

🏆
ASME CIE Best Ph.D. Dissertation Award 2024 ASME Computers and Information in Engineering Division
🥇
Best Paper Award – ACM/SIGAPP SAC 2024, AI & Agents Theme "Graph Learning-based Fleet Scheduling for Urban Air Mobility under Operational Constraints, Varying Demand & Uncertainties"
🎖️
Outstanding Coordination Paper Nomination – IEEE ICRA 2022 "Learning Scalable Policies over Graphs for Multi-Robot Task Allocation using Capsule Attention Networks"
✈️
IEEE RAS Travel Grant – ICRA 2023 IEEE Robotics and Automation Society
✈️
IEEE RAS Member Supported Program Grant – IROS 2023 IEEE Robotics and Automation Society

Collaborations & Service

Academic & Industry Collaborations
REIL Lab, University of Connecticut
Joint Postdoc · 2024–2025
Cross-institutional postdoctoral appointment bridging academic research in XAI and prognostics with industry deployment at Percev LLC. Joint work on bearing fault detection, survival analysis, and state estimation for high-rate dynamical systems.
University of Texas at Dallas
Research Collaboration · Prof. Jie Zhang & Prof. Yulia Gel
Joint work on Graph Reinforcement Learning for power distribution network reconfiguration and outage management, culminating in a Nature Communications publication (2024) and multiple IEEE conference papers.
University of South Carolina / Iowa State University
Research Collaboration · Prof. Simon Laflamme group
Collaborated on probabilistic state estimation for high-rate dynamic systems using Topological Data Analysis and ML pipelines, published in Mechanical Systems and Signal Processing (2025) and DDDAS 2024.
ADAMS Lab, University at Buffalo
Ph.D. Research · Prof. Souma Chowdhury
Developed Graph RL frameworks for multi-robot planning, UAM fleet scheduling, and network optimization. Funded by ONR AI/ML (N00014-21-1-2530) and NSF CAREER (CMMI 204802). Collaborated with UB CSE (Prof. Karthik Dantu) on UAM vertiport scheduling.
Percev LLC & Grace Technologies
Industry Partnership · 2024–2025
Translated academic research on bearing fault detection and survival analysis into a commercial product for industrial customers. Built and deployed end-to-end ML pipelines from raw vibration data to customer-facing diagnostics UI.
Florida Institute for Human & Machine Cognition (IHMC)
Invited Talk · March 2025
"Cyber-Physical Systems for the Future: Trust, Adaptability, and Intelligence" — invited presentation at IHMC, Pensacola, FL, covering AI-based fault detection, explainability, and autonomous decision-making for physical systems.
Mentoring
Mohammad Mundiwala
Ph.D. · Explainable AI for PHM of rotating machines · UConn
Yang Kang Chua
Ph.D. · State estimation for high-rate dynamical systems using TDA · UConn
Nathan Maurer
M.S. · Multi-agent collective transport using GNN & bipartite graph matching · UB
Jhoel Witter
M.S. · UAM eVTOL scheduling using Graph RL · UB
Wenyuan Li
M.S. · Multi-agent collective transport via auction-based methods · UB
Aditya Divyakanth Bhatt
M.S. · Multi-robot search using hybrid imitation learning · UB
Thanh Mai & Khoi Pham
Interns · Analytics & ML · Percev LLC, Summer 2025
Brian Smyth & Ibtida Ahmed
Undergrads · Simulation environments for multi-robot & power network applications · UB
Teaching
Course Design
Co-designed Learning for Autonomous Systems (MAE 600 / CSE 610, Spring 2022), a graduate-level course at University at Buffalo — developed curriculum, lecture slides, and course projects; mentored students throughout.
Guest Lecture
Learning for Autonomous Systems (MAE 600 / CSE 610, Spring 2024) — invited lecture on Graph Reinforcement Learning for Robotics.
Teaching Assistant
Engineering Computations (EAS 230, Spring 2020) · Manufacturing Automation (MAE 564, Fall 2020) · Intermediate Dynamics (MAE 345, Spring 2021)
Professional Service
Program Committee
ACM/SIGAPP Symposium on Applied Computing (SAC) – IRMAS track, 2025
Session Chair
AIAA ISFA 2024 (4 sessions) · AIAA AVIATION 2024 (Physics-Informed ML, Emerging Methods) · ASME IDETC 2024
Peer Review
114+ reviews across NeurIPS (2022–2025), ICRA (2021–2024), IROS (2022–2025), ICML 2025, ICLR 2024, ICAPS 2024, and journals including Nature Scientific Reports, IEEE TRO, IJRR, RA-L, Robotics & Autonomous Systems, and more.
Judging
AIAA Graduate Student Awards, 2025
Memberships
IEEE · ASME · AIAA · ACM
Media Coverage
Tech Explorist Science Daily American Public Power Association Nature Communications – Editors' Highlights UN Office for Disaster Risk Reduction (PreventionWeb) Futurity Design Products & Applications