Widener University
Assistant Professor of Supply Chain & Data Science
AI/ML systems · funded initiatives · teaching · national serviceAminKeramati, PhD
Current
Assistant Professor of Supply Chain & Data Science
Widener UniversityI engineer artificial-intelligence, data, and machine-learning systems for consequential transportation and infrastructure decisions—from autonomous-vehicle and LiDAR pipelines to survival models, agentic LLM workflows, safety validation, and agency-ready decision tools.
01 / EXPERIENCE & EDUCATION
Experience & education
Experience
UGPTI / NDSU
Graduate Research Assistant
Federal safety systems · survival ML · optimization · GISEducation
PhD
Transportation & Logistics
North Dakota State University · GPA 3.94MS
Industrial Management
Financial Engineering · optimizationBS
Industrial Engineering
Systems, operations, and quantitative analysis10+years in ML and safety analytics
20+AI, ML & supply-chain courses, workshops, and presentations designed and taught
>$200KU.S. DOT, academic & research initiatives
2research and GIS awards
02 / TECHNICAL EXPERTISE
Technical expertise
Amin’s capabilities span artificial-intelligence and ML engineering, programming, data pipelines, statistical modeling, optimization, LiDAR, GIS, and safety-system design for autonomous vehicles, railway, and highway applications.
01Ingest
02Validate
03Transform
04Engineer features
05Model
06Evaluate
07Deliver
AI & machine-learning engineering
End-to-end artificial-intelligence systems spanning time-to-event and competing-risk models, monotonic neural networks, random survival forests, gradient boosting, graph neural networks, feature engineering, calibration, deployment-oriented evaluation, RAG, and agentic LLM workflows.
PythonPyTorchscikit-learnDGCNNRAGAgentic AI
Data engineering
Source integration, cleaning, validation, transformation, temporal alignment, spatial joins, model-ready feature pipelines, database work, and reproducible analytical workflows.
PythonSQLMySQLR
Data mining & analytics
Severity-specific signals, nonlinear relationships, temporal behavior, spatial patterns, variable importance, scenario analysis, optimization, and decision-focused interpretation.
PandasNumPyGAMSTableau
Geospatial, LiDAR & decision systems
AI-driven LiDAR point-cloud pipelines, transportation networks, GIS integration, agency dashboards, enterprise systems, and analytical interfaces that make complex models usable.
LiDAR AIArcGISR ShinySAP HANASAP ERP
Safety systems design & engineering
AI-enabled safety-system architecture for autonomous vehicles, railway, and highway environments—connecting hazard identification, conflict scenarios, time-to-event risk, countermeasure logic, LLM guardrails, validation, and decision support.
AV safetyRailway safetyHighway safetyAI assuranceSystem validation
03 / SELECTED PROJECTS
Selected projects
In-progress and completed AI, machine-learning, data-science, optimization, and transportation-safety systems. Each case connects the problem, data pipeline, method, validation, numerical result, and practical decision.
In progress
Two active AI and safety-engineering programs
P01In development · Competing-risks survival analysis
Time-to-conflict — AV safety survival analysis
A competing-risks analysis of 22,898 Waymo vehicle-scenarios showing fewer rear-end conflicts and more lane-change conflicts for AVs—opposing effects that a pooled model cancels into ‘no difference.’
- Raw trajectories
- Validated conflicts
- Competing risks
- Python / R replication
- Waymo Open Motion
- Competing risks
- Cause-specific Cox
- Computational geometry
- SOTIF
9,044conflicts extracted
P02In development · Agentic AI safety engineering
Grade-crossing safety intelligence
A survival-ML, RAG, and agentic LLM workflow that converts highway–rail crash risk into traceable interventions, audits its own evidence, and refuses recommendations when support is insufficient.
- Crossing data
- Survival engine
- Grounded retrieval
- Audited decision
- AI engineering
- Agentic LLMs
- RAG
- FAISS
- Safety guardrails
3,310crossing records
Completed projects
Nine practical systems · newest first
C01March 2026
2–50%discrimination improvement
AI/ML engineering · Safety modeling
Monotonic neural crash severity
Explore the technical caseEngineered four monotonic neural architectures that learn when a crash may occur and which severity may emerge, while preserving physically defensible risk behavior across competing outcomes.
- 29-year records
- Survival tensors
- Monotonic neural heads
- Calibrated risk
My roleProject Lead · Lead Developer · First Author
- AI engineering
- PyTorch
- Monotonic networks
- Competing risks
C02May 2025
99.32%lengthwise correctness
AI perception · LiDAR engineering
LiDAR rail extraction
Explore the technical caseEngineered an AI-driven LiDAR pipeline that combines local geometric filters with a dynamic graph neural network to reconstruct rail geometry from sparse, noisy point clouds without relying on sensor configuration or fixed global rail features.
- Sparse point cloud
- Local filtering
- DGCNN
- Rail geometry
My roleKey Investigator · Data Contributor · Coauthor
- AI engineering
- LiDAR
- DGCNN
- Point-cloud engineering
C03March 2025
4.73%crossings classified high-risk
Data science · Decision engineering
AHP hazard ranking
Explore the technical caseBuilt a transparent hazard-ranking engine that combines expert safety priorities with severity-specific crash probabilities, converting model output into an actionable statewide screening system.
- Crash histories
- Severity probabilities
- Expert weighting
- Priority levels
My roleMethodology Lead · Model Developer · First Author
- AHP
- Risk index
- Competing risks
C04February 2023
>$200Kfunded initiatives
AI-enabled safety engineering · Decision systems
DOT safety support system
Explore the technical caseLed the technical integration of crash-risk modeling, countermeasure analysis, GIS prioritization, and an interactive R Shiny application into one agency-facing safety workflow.
- Agency records
- Model + GIS integration
- R Shiny
- Safety scenarios
My roleTechnical Project Lead · Lead Developer · Key Investigator
- AI/ML
- R Shiny
- GIS
- Decision support
C05August 2021
29 yearsnetwork history analyzed
Safety analytics · Causal comparison
Countermeasure effectiveness
Explore the technical caseQuantified how the safety value of gates, audible warnings, flashing lights, and stop signs changes with the controls already present at a crossing.
- 29-year network
- Control-pair encoding
- Competing risks
- Effect estimates
My roleProject Lead · Model Developer · First Author
- Cox model
- Marginal effects
- CMF engineering
C06August 2020
3,310public crossings modeled
Statistical ML · Rare-event safety
Competing-risk safety model
Explore the technical caseDesigned a single competing-risk framework that estimates whether a crash occurs and which severity follows, using all crossing records rather than discarding sites with no observed crash.
- Full inventory
- Censoring + event codes
- Cause-specific Cox
- Cumulative incidence
My roleProject Lead · Model Developer · First Author
- Survival analysis
- Cox regression
- Risk ranking
C07July 2020
<2%fatal-risk prediction error
AI/ML engineering · Explainable risk
Random survival forest
Explore the technical caseBuilt an open-source random survival forest that learns nonlinear, severity-specific relationships and ranks the contributors that matter most over a 29-year risk horizon.
- Crossing records
- Survival forest
- OOB validation
- Severity ranking
My roleProject Lead · Model Developer · First Author
- AI/ML
- Random survival forest
- VIMP
- Open source
C08February 2020
16–34%transport-cost reduction
Optimization · Sustainable logistics
Forest logistics optimization
Explore the technical caseEngineered a statewide network model to test how forest-road maintenance policies reshape mill access, log-truck routing, transportation cost, energy use, and carbon emissions.
- Road + mill data
- GIS network
- Policy scenarios
- Cost + emissions
My roleOptimization Lead · Model Developer · First Author
- Network optimization
- ArcGIS
- Scenario engineering
C09January 2020
20.9%30-year crash probability with 3 tracks
Geospatial data science · Safety engineering
Geometric crossing safety
Explore the technical caseEngineered continuous geometric features from aligned GIS layers, then modeled how intersection distance, crossing angle, road lanes, and main tracks change long-term crash occurrence and severity.
- GIS + FRA records
- Spatial alignment
- Competing-risk model
- 30-year risk curves
My roleLead Model Developer · Geospatial Data Engineer · First Author
- GIS feature engineering
- Competing risks
- Cumulative incidence
04 / SELECTED DESIGNED TOOLS & APPLICATIONS
Selected Designed Tools & Applications
Interactive products Amin designed and developed to turn transportation-safety methods, AI decision logic, and complex analytical workflows into usable experiments and interfaces.
A01 Live product demo
Manual road-user control21-second loop · begins after the original three-second delay
Designed & built · AV safety engineering
AV Traffic Conflict Simulator
An interactive browser laboratory for configuring AV–road-user encounters, measuring TTC and PET, inspecting conflict-zone occupancy, and stress-testing AV behavior through human-controlled scenarios.
- Configure
- Simulate
- Measure
- Inspect
My roleProduct Designer · Lead Developer
- JavaScript
- TTC / PET
- AV decisioning
- Human-in-the-loop
- Responsive web app
A02 Full video demo
Multi-origin place findingComplete interface recording · full frame preserved
Designed & built · Geospatial decision application
THERE
An R Shiny location-choice application that searches candidate places around multiple origins, compares travel time by mode and departure settings, and maps the strongest shared destinations for personal or logistics decisions.
- Add origins
- Find places
- Compare time
- Map options
My roleProduct Designer · Lead Developer
- R Shiny
- Google Places
- Google Directions
- Leaflet
- Travel-time analysis
A03 Full video demo
Crossing-risk scenario analysisComplete application recording · full frame preserved
Designed & built · Railway safety decision support
Crash Predictor at HRGCs
An R Shiny safety application that applies a competing-risk survival model to compare highway–rail crossing controls and geometry, then visualizes crash-occurrence and severity likelihood across a 29-year horizon.
- Set crossing
- Compare controls
- Predict risk
- Review change
My roleTechnical Project Lead · Lead Developer · Key Researcher
- R Shiny
- Competing risks
- Cox survival model
- HRGC safety
- Countermeasure analysis
05 / SELECTED PUBLICATIONS
Selected publications
Amin Keramati’s peer-reviewed work in AI and machine learning, transportation safety, highway–rail systems, LiDAR, statistical modeling, optimization, and decision analytics.
Full Google Scholar profileTime-to-event crash severity prediction at highway-rail grade crossings with monotonic neural networks
Accident Analysis & Prevention · A* / Q1
Evaluating the impact of road-user actions on crash severity at highway-rail grade crossings
Infrastructures · Q2
Designing football matchday experiences: influence of stadium touchpoints on fan evaluation
Managing Sport and Leisure
Evaluating crash severity using an Analytic Hierarchy Process-based hazard index
Accident Analysis & Prevention · A* / Q1
A hybrid local-feature-based approach for automated rail extraction from LiDAR data
Transportation Research Record · Q2
Optimal siting of biorefineries for a switchgrass-based bioethanol supply chain
Logistics · Q2
Effectiveness of safety countermeasures at highway-rail crossings using competing risks
Journal of Safety Research · A / Q1
Crash severity analysis at highway-rail grade crossings: the random survival forests method
Accident Analysis & Prevention · A* / Q1
Geometric effect analysis of highway-rail grade-crossing safety performance
Accident Analysis & Prevention · A* / Q1
Simultaneous analysis of crash frequency and severity using competing risks
Journal of Advanced Transportation · A / Q2
Gradient-boosting crash prediction for highway-rail grade crossings
Journal of Advanced Transportation · Q2
Forest-road maintenance policy effects on log transportation, routing, and carbon emissions
Journal of Transportation Engineering, Part A · Q2
06 / TEACHING
Teaching experience
Assistant Professor, Widener University · PA, USA2021—Present
Amin’s graduate and undergraduate teaching covers artificial intelligence, machine learning, database systems, data analytics, ERP/SAP, operations, supply chains, project management, and transportation optimization through real datasets, working code, and decision-focused projects.
Widener UniversitySchool of Business Administration · PA, USA
9 courses01 · Spring 2022—PresentUndergraduate
Supply Chain Management SCM 460
Supply-chain principles, ERP and SAP applications, and spatial supply-chain analysis using ArcGIS Pro, Python, and SQL.
02 · Spring 2022—PresentUndergraduate
Operations Management SCM 352
Core operations concepts, quantitative decision-making, process analysis, and operational planning.
03 · Fall 2021—PresentGraduate
Database Systems IS 610
Relational design, SQL, normalization, data warehousing, MySQL, Tableau dashboards, and SAS-based data management.
04 · Fall 2021—PresentGraduate
Information Systems and Data Analysis BUS 615
Information-systems strategy, business intelligence, analytics, decision modeling, data mining, Big Data, cloud technologies, SaaS, and management of AI.
05 · Summer 2025Graduate
Project Management MGT 680
Project planning, scheduling, resource allocation, risk management, and project-performance evaluation.
06 · Summer 2025Graduate
AI Essentials MGT 688
Machine-learning fundamentals, hands-on model development and evaluation, AI applications, and applied business cases.
07 · Fall 2022Undergraduate
Management Information Systems MIS 290
Principles of management information systems and their application to business decision-making.
08 · Fall 2021Undergraduate
ERP Systems and Workflow Management MIS 430
ERP fundamentals, business-workflow integration, and a hands-on introduction to SAP.
09 · Course taughtUndergraduate
Database Systems MIS 358
Relational design, SQL, data warehousing, MySQL, and business-dashboard development using Tableau.
North Dakota State UniversityCollege of Business · Fargo, ND
Teaching presenter10 · Summer 2016Graduate · Teaching Presenter
Modeling for Transportation & Logistics Decision Analysis TL 831
Transportation optimization, including shortest-path, traveling-salesperson, and vehicle-routing problems.
11 · Fall 2015Graduate · Teaching Presenter
Modeling for Logistics Research TL 811
Rail planning, freight routing and scheduling, LP/MIP optimization, service-network design, time-space networks, and railcar routing.
Graduate + undergraduateAI + machine learningProject-centeredIndustry datasetsOptimizationTechnical-to-practical translation
07 / ABOUT
About Amin Keramati
I’m Amin Keramati, an Assistant Professor of Supply Chain & Data Science at Widener University and an AI/ML engineer focused on making safety-critical transportation systems more reliable and useful.
My work brings together artificial-intelligence engineering, data engineering, survival analysis, deep learning, agentic LLMs, RAG, LiDAR, optimization, GIS, and safety decision-system design. I enjoy working across the full path—from source data and mathematical formulation to code, validation, AI assurance, and the interface an engineer or agency analyst ultimately uses.
I also contribute to Transportation Research Board committees and national panels focused on rail safety, crash-modification factors, intelligent transportation systems, and responsible use of AI at state DOTs.
AI engineeringTransportation safetyMachine learningAgentic LLMsLiDARData pipelinesOptimizationDecision systems
Profile documents
Curriculum VitaeUpdated academic record · PDFResumeML and AI engineering · PDFContact optionsProfessional and technical inquiriesDetailed academic record or concise industry-focused experience.
08 / MEDIA
Media
Selected conversations and public features connecting transportation safety, artificial intelligence, applied analytics, and decision-making beyond the technical paper.
PodcastA Widener University Podcast
Marketing Research, AI, and the Fan Experience in Major League Sports
Greg Potter speaks with professors Brian Larson and Amin Keramati and marketing student Ben Miller about their Philadelphia Union fan-experience research, its international reach, and how human-centered marketing, AI, and facial recognition can improve experience, efficiency, and decision-making.
More informationWalletHub · Expert feature
Delaware Car Insurance: Ask the Experts
Amin contributes a transportation-safety and risk perspective to WalletHub’s expert feature on Delaware car insurance and the factors consumers should consider when evaluating coverage and cost.
More information
University of Nebraska–Lincoln · Seminar
A Crash Severity Analysis at Highway-Rail Grade Crossings: The Random Survival Forest Method
A Transportation Engineering Seminar Series presentation by Dr. Pan Lu and Amin Keramati on competing-risk Random Survival Forest modeling for crash-severity analysis at highway-rail grade crossings.
September 20, 2024College of EngineeringNebraska Transportation Center
More information09 / CONTACT
Contact Amin Keramati
Contact information and profile documents for AI/ML engineering, data science, transportation safety, teaching, and technical collaboration.
Email Amin