Martin Walter Ampah, Applied ML Researcher & Engineer: Predict. Secure. Trust.

(  Research focus  )

What I work on.

Three pillars

I build AI that predicts well, fails safely, and knows when to defer to a person when necessary.

  1. (001)

    AI prediction

    Forecasting failures and risk from sensor and market data.

  2. (002)

    AI security

    Stress-testing models and pipelines so they fail safely, not silently.

  3. (003)

    Trustworthy AI

    Calibrated confidence and conformal guarantees that flag when a human should decide.

(  Selected work  )

Research that ships.

02 projects
[001]

Safety-Aware Predictive Maintenance

MSc research · Kumamoto University

  • Python
  • Conformal prediction
  • Calibration
  1. 01Sensor data
  2. 02Training
  3. 03Calibration
  4. 04Conformal sets
  5. 05Trust or defer

Predicts rare equipment failures, then decides when the model can be trusted and when to defer to a human.

[002]

Market Prediction & Risk Analysis

Independent research

  • Python
  • XGBoost
  • LSTM
  1. 01Market data
  2. 02Features
  3. 03Validation
  4. 04Forecasting
  5. 05Risk analysis

Forecasts equities, crypto, FX, and commodities through tested, reproducible pipelines with built-in anomaly checks.

(  Experience  )

Six years in production.

Since 2019 · Remote
  • Built Python and TypeScript automation for a platform serving tens of millions.
  • Internal tooling cut repetitive effort by ~40%; root-caused 100+ issues.
  • Automation and API validation for AI Chat, messaging, and payments.
  • Re-architected the test framework, cutting run time by ~30%.
  • Integration tooling for FinTech authentication and financial workflows.
  • Strengthened CI/CD and production diagnostics.
  • Internal tools and reusable components shared across projects.
  • Monitoring for system health and quality metrics.
(  Toolkit  )

Built to be measured.

Three disciplines

[01]Prediction

  • Python
  • XGBoost
  • LSTM
  • Time-series forecasting
  • Predictive maintenance
  • pandas

[02]Trust & security

  • Conformal prediction
  • Probability calibration
  • Uncertainty quantification
  • Explainable AI
  • Anomaly detection
  • Integration testing

[03]Engineering

  • REST APIs
  • Docker
  • CI/CD
  • GitHub Actions
  • Linux
  • SQL
(  Education  )
2026Kumamoto, Japan

MSc Computer Science

Kumamoto University

2019Berekuso, Ghana

BSc Electrical Engineering

Ashesi University