This project was my final project for ALY6130: Risk Management for Analytics.
I used NVIDIA as the case company and built a structured risk management analysis around one main question:
How can NVIDIA identify, prioritize, and respond to major business risks in a practical way?
This GitHub version is my cleaned public version. I kept the final report, the final briefing slides, selected visuals, one notebook, and a small archive of original project files.
Main links:
NVIDIA has strong growth in GPUs and AI, but it also faces several risks that could hurt operations and revenue.
The biggest concerns in this project were:
- dependence on TSMC
- export restrictions
- competition from AMD and Intel
- geopolitical tensions around U.S.-China relations and Taiwan
- lower-cost AI competition
- market valuation volatility
The final version focused most on two high-priority risks:
- Supply Chain Disruptions
- Geopolitical Tensions
The course project was built step by step across multiple modules.
The final assignment asked for a full risk management project that included:
- risk identification
- qualitative and quantitative assessment
- KRI
- scenario modeling
- treatment and response planning
- communication and continuous monitoring
I kept that overall structure in the final version of this repo.
This project is not built around a large downloaded dataset.
The main inputs were:
- public company filings
- public business and news sources
- manually built risk scores
- project working Excel files
- manually defined scenario assumptions for six-quarter sales
- simulated revenue outputs
Supporting project working files are kept in:
See the full note here:
My workflow for this project was:
- Identify major NVIDIA business risks
- Build a six-risk register
- Score risks by chance and impact
- Use a heat map to narrow the focus
- Build an influence diagram and scenario tree
- Summarize six-quarter revenue impact with Monte Carlo analysis
- Translate the results into mitigation, KRI, and response planning
The final project used six risks:
- Supply Chain Disruptions
- Export Restrictions
- Strong Competition
- Geopolitical Tensions
- Emergence of Cost-Effective AI Models
- Market Valuation Volatility
The final scoring highlighted two high-priority risks:
- Supply Chain Disruptions
- Geopolitical Tensions
In this GitHub version, the risk register and scoring details are mainly preserved through the archive working files and the selected figures shown here.
After scoring the risks, I used a heat map to show which risks were both high chance and high impact.
That visual helped confirm that the project should focus more on:
- supply chain risk
- geopolitical risk
In the public repo, that heat map is kept as a supporting figure rather than rebuilt step by step in the notebook.
In the later project stage, I focused more on Geopolitical Tensions as the main modeling case.
The influence diagram connected geopolitical instability to:
- trade restrictions
- manufacturing delays
- supply chain disruptions
- sales impact
The updated scenario tree translated those ideas into three main paths:
- regulatory compliance
- manufacturing delays
- trade restrictions
The notebook in this repo is:
The notebook focuses on the main public analysis logic:
- an earlier scenario comparison stage
- updated Monte Carlo simulation
- updated CDF comparison
- the later updated three-path scenario comparison
def monte_carlo_simulation(sales_data, simulations=10000, quarters=6):
results = []
for _ in range(simulations):
total_sales = 0
for _ in range(quarters):
scenario = np.random.choice(
["Calm", "Moderate", "Intense"],
p=[0.25, 0.50, 0.25]
)
sales = np.random.choice(sales_data[scenario])
total_sales += sales
results.append(total_sales)
return resultsThis part mattered because it turned the scenario assumptions into a simple six-quarter revenue impact comparison.
In this repo, the notebook is mainly the simulation part of the project. The risk register, scoring setup, and heat map are preserved through archive/source artifacts and selected visuals instead of being fully recreated in notebook code.
The notebook also keeps an earlier comparison stage and a later updated three-path stage in the same file. The later updated stage is the main public-facing result used in the final comparison.
The simulation values and scenario probabilities are manually defined project assumptions. The notebook does not set a fixed random seed, so reruns may vary slightly.
The updated final version compared three paths:
- Regulatory Compliance
- Manufacturing Delays
- Trade Restrictions
My main read from the final comparison was:
- Regulatory Compliance had a narrower and safer range
- Manufacturing Delays showed more uncertainty
- Trade Restrictions showed the biggest downside risk
My main conclusion was that NVIDIA’s biggest risks were not only general business threats.
The more useful point was that:
- the risks could be structured and scored
- the top risks could be translated into scenario logic
- the analysis could support practical action ideas like supplier diversification, fab expansion, KRI thresholds, and faster response planning
So the project became more like business risk decision support, not just a class report.
A simple way I would explain this project is:
This was my individual risk management analytics project on NVIDIA.
I started with a six-risk register and used scoring plus a heat map to narrow the focus to supply chain and geopolitical risk.
Then I used scenario modeling and Monte Carlo analysis to show how those risks could affect revenue over the next six quarters, and I connected the results to mitigation planning and KRI design.