ASML-System-Integration-and-Software-Testing-Engineer

Role: System Integration & Software Testing Intern (Part-Time: 24 hrs/wk)

When: Jan 2024 – Dec 2024 · Location: Taipei, Taiwan

Disclaimer: This portfolio uses sanitized descriptions and synthetic examples only. No proprietary code, data, screenshots, internal IDs, or configurations from ASML are included.

---

Projects at a Glance

Project	Problem	What I Built	Result	Stack
P1. YieldStar Log Quality Assurance Automated Pipeline	Noisy and multi-source logs slowed triage	Python validator packaged as a one-click tool with schema/regex checks, caching, retry	Processing time ↓ ~90%	Python, Pandas (text analytics), Pydantic/JSON Schema (validation), Pytest/Hypothesis (tests), Tenacity (retry), Redis (cache)
P2. Robot Framework API Test Suites & CI Quality Gates	Post-merge defects surfaced late	Robot Framework API tests and GitHub Actions blocking PRs; JUnit reports for inspection	Post-merge defects ↓ ~80%	Robot Framework, GitHub Actions, JUnit XML, pytest/coverage, pre-commit, Docker
P3. React Log Viewer with Anomaly Flags	Slow handoff from ops to investigation	Web application with anomaly flags via threshold & timestamp checks	Handoff-to-investigation time ↓ ~60%	React, TypeScript, React Query, Docker, REST API (Flask/FastAPI)
P4. PowerBI Ops Dashboards	Limited visibility into yieldstar machine logs health	Power BI dashboards with daily refresh, health checks, and runbooks	Faster daily reviews that support internal and external projects	Power BI, DAX, Power Query (M), Scheduled Refresh, REST/CSV connectors, SQL

---

P1 — YieldStar Log Quality-Assurance Automated Pipeline

Problem: Engineers manually parsed logs from multiple sources, delaying defect discovery.

Solution: A validator packaged as a one-click tool that normalizes records, enforces schemas, flags anomalies, and caches results.

Highlights:

• Validation & normalization: Utilize schema and regex checks, and normalize timestamps and fields.
• Resilience: Caching and idempotent retries for unstable inputs.
• Tooling & tests: Command-line tool (CLI) and synthetic fixtures for unit tests.

Impact:

• Processing time reduced ~90%
• Packaged as a one-click CLI and adopted by the Software Testing team, now used daily in log triage

Pipeline (simplified)

flowchart LR
  A["Raw Logs (synthetic)"]
  B["Validator (schema + regex)"]
  C["Cache / Retry"]
  D["Validated Records"]
  E["Results API"]
  F["Downstream Consumers"]

  A --> B
  B --> C
  B --> D
  D --> E
  E --> F

Loading

---

P2 — Robot Framework API Test Suites & CI Quality Gates

Problem: Yieldstar Machine Logs post-merge defects surfaced late and occasionally reached main.

Solution: Robot Framework API test suites run in GitHub Actions; failing checks block PR merges. Test outputs are surfaced as JUnit XML in the PR UI, with coverage reported.

Highlights:

• API checks: Ensure endpoints return correct status codes (e.g., 200/400), JSON matches the expected schema, and bad inputs produce clear errors.
• Consistent CI: Run automatic pre-commit checks and execute CI inside Docker so tests behave the same on every machine.
• Coverage: Use pytest --cov to report how much of the code is exercised by tests.

Impact:

• Post-merge defects ↓ ~80%

Pipeline (simplified)

flowchart LR
  A["YieldStar Log Pull Request"] --> B["GitHub Actions"]
  B --> C["Robot Framework Suites"]
  C --> D["JUnit XML + Coverage"]
  D --> E["PR Status Checks"]
  E -->|fail| F["Block Merge"]
  E -->|pass| G["Merge to Main"]

Loading

---

P3 — React Log Viewer with Anomaly Flags

Problem: Ops needed faster triage and clearer signals to move from logs to investigation state.

Solution: A React web app that highlights anomalies via threshold and timestamp checks, supports filters and search, and links directly to the source record for follow-up.

Highlights:

• See issues at a glance: Show row-level anomaly badges and time markers.
• Work faster: Filter by module, severity, and time window and instant search.
• One click to investigate: Build deep links to the exact record.
• Smooth data loading: React Query handles fetch, caching, and retries; TypeScript for safer user interface code.

Impact:

• Handoff-to-investigation time ↓ ~60%

UI Flow (simplified)

flowchart LR
  A[Validated Records API] --> B[React Viewer]
  B --> C[Anomaly Flags]
  B --> D[Filters & Search]
  B --> E[Deep Links to Investigation]

Loading

---

P4 — Power BI Ops Dashboards

Problem: Daily job health and anomaly trends were scattered across tools, making status reviews slow and inconsistent.

Solution: A set of Power BI dashboards with daily refresh, health checks, and runbooks so ops can review status in minutes and know exactly what to do next.

Highlights:

• Pipeline health: Show success rate, latency, retry counts, recent failures, and so on.
• Anomaly overview: Flagged records by type, time, and module with drill-through and filters.
• Runbooks: Links to step-by-step fixes, owners, and escalation paths.
• Reliable data flow: Build Power Query (M) for shaping, scheduled refresh, and REST/CSV connectors bring in the sources.

Impact:

• Faster daily reviews and clearer ownership and escalation during incidents
• Used across domestic and international teams for daily status reviews and incident follow-ups

Data Model (simplified)

flowchart LR
  A["Validated Records (synthetic)"] --> B["Power Query (M)"]
  B --> C["Model (DAX)"]
  C --> D["Dashboards & Alerts"]

Loading