Command-centre intelligence
Turn operational workflows into system behaviour and acceptance criteria.
Requirements · KPI logic · validationAI Nassar · AI Agent
I'm AI Nassar, your AI agent. Ask me about my work, research, and projects.
RESEARCH · FOUR-CITY BENCHMARK
Better forecasts don’t always lead to better decisions.
Four cities. 30.2 million retained records. One test: forecast to simulated allocation.
The question
When a forecast improves, does the decision improve with it?
Evidence at scale
36.8M → 30.2M
Municipal 311 reports across New York, Chicago, San Francisco, and Austin · 2020–2025.
- records acquired
- 36.8M
- records retained
- 30.2M
- cities
- 04
The method
Build the benchmark. Hold out the future. Test the decision.
- 01
Harmonise
Four reporting systems. One shared service schema.
- 02
Forecast
Held-out temporal evaluation with leakage controls.
- 03
Allocate
Three capacity bands in a controlled simulation.
- 04
Measure
Unresolved demand against uniform allocation.
The finding
Accuracy improved. Decision value still depended on how the forecast was represented.
Capacity, objective, and uncertainty shaped what the allocation rule could distinguish.
Austin10.2%
San Francisco10.4%
New York11.9%
Chicago15.6%
92–100%of allocation steps tied under the tested point-forecast objective
ρ 0.88–0.99Spearman rank correlation between model and decision rankings across tested city–capacity cells
Scarce18–23%
Moderate38–70%
More generous60–93%
Reproducible benchmark · controlled allocation simulation · not a live municipal deployment
EARLIER STUDY · NEW YORK CITY
Forecast gains translate differently with capacity.
An 11–15% forecasting gain translated into three different simulated allocation gains.
0.4%Scarce capacity
1.9%Baseline setting
14.5%More generous capacity
CURRENT WORK
Turn operational needs into testable systems.
From requirements and KPI logic to AI-enabled workflows and validation.
FROM NEED TO DELIVERY
FROM NEED TO DELIVERY
- 01
Operational need
Define the workflow, constraints, and decision.
- 02
Requirements
Turn the need into system behaviour a technical team can build.
- 03
Data & KPI logic
Define data fields, KPI logic, thresholds, and exceptions.
- 04
AI & systems
Connect the workflow to analytics, AI agents, or digital twins.
- 05
Validation
Set acceptance criteria and support user testing.
- 06
Operational value
Confirm the result supports a clear next action.
WORK IN PRACTICE
Where operational needs become technical decisions.
KPI digitisation
Define data, KPI, and display requirements across 34 operational assets.
34 operational assetsTicketing & automation
Define trigger thresholds, map procedures, and support user acceptance testing.
Cross-functional deliveryBuilt to be tested in the workflow.
EDUCATION
Built on engineering. Specialised in AI.
Computer engineering built the foundation. Advanced AI became the specialisation.
2017–2022FOUNDATION
Computer Engineering
A systems foundation spanning hardware, software, computing, and digital design at Umm Al-Qura University.
Bachelor of Computer Engineering · 178 credit hours2023–2025SPECIALISATION
Master of Artificial Intelligence
Advanced AI at the University of Technology Sydney, with a sub-major in Natural Language Processing.
GPA 6.13/7 · WAM 82.44 · 5 High Distinctions
The foundation carries forward.
Engineering and AI now meet in applied work.
CONTACT
Let’s talk.
For opportunities, research, or a useful conversation, choose what works best for you.