Haoming Koo.

Build and ship production ML systems end-to-end: inference APIs, containerised deployment, MLOps pipelines, and frontend UIs across vision, sequence, and NLP domains.

• Delivered production-ready pipelines using GitLab CI/CD, Docker, and Google Cloud – including backend inference services and frontend UIs
• Built vision models for image classification, object detection, and quality checks
• Developed sequence modelling systems for time-series prediction
• Implemented containerised deployment, model lifecycle management, and MLOps automation
• Contributing to an industry AI solution with AMD through AI Singapore's 100E programme
• Separately built 8 public portfolio applications integrating ML models with cloud-hosted dashboards (kooexperience.com)
• Expanding into NLP, LLM architectures, LangGraph-style orchestration, observability, and agentic AI workflows

A year focused on hands-on AI product building after Micron: public applications that combine models, APIs, UI, validation, and deployment.

• Completed Full Stack Development with AI (NUS x Emeritus, 24 weeks) – React, Flask, MongoDB, SQL, Node.js, REST APIs, AI integration
• Completed GA100 Generative AI (Heicoders Academy, WSQ) – prompt engineering, RAG, AI agents, workflow automation (n8n), MCP
• Built public portfolio applications deployed at kooexperience.com: LLM workflows, agent systems, vision pipelines, and RAG
• Shipped a production MCP server (Japan in Seasons) with 17 read-only tools and 1,700+ seasonal travel places
• PMP certification

Led global AI-enabled transformation and cross-site program delivery across four fabs (Singapore, Boise, Hiroshima, Taichung). Managed 6 direct and up to 9 indirect engineers with a USD 9M operational portfolio.

• Designed and executed a 5-year autonomous conversion roadmap using 3M strategy – $50M+ in realised savings
• Unified fragmented process platforms into a single AI-integrated ecosystem – standardised 80% of workflows across 4 countries
• Co-led a flagship low-power DRAM yield ramp – maturity pulled in by 7 months on a multi-quarter mobile-DRAM program
• Co-led a high-bandwidth-memory yield program – supported Micron's first HBM market entry
• Modernised infrastructure from Hadoop to Google Cloud Platform (GCP)
• Built and deployed a production ML risk-scoring system with real-time pipeline orchestration – 1–2% yield improvement at production scale
• Designed and deployed a production GenAI SOP workflow with retrieval, prompt orchestration, and validation layers – 80–90% answer relevance, ~15 min/query deflection, ~2 weeks faster onboarding, scaled to ~3,000 engineers
• Scaled RPA (UiPath/Power Automate) for targeting-lot workflow – 100% adoption; mentored ~30 engineers
• Built reliability/observability: pipeline health dashboards, canary/shadow runs, rollback runbooks – near-zero ML-pipeline outages
• Drove SEMulator3D rollout: accelerated validation by 2 weeks, >$10M in time-to-market gains
• Conducted red-team UAT simulations, reducing post-deployment issues by 25%

Transitioned from equipment engineering to data-driven automation – developing ML-enabled defect prediction and control systems.

• Built ResNet-50 computer vision model for wafer/chamber misplacement detection – earned Micron Innovation Award
• Scaled AI-based OCR defect detection system fab-wide – 4–5% yield improvement
• Reduced scanner wafer rejection to <0.5% and MA excursion impact to <0.1%
• Developed anomaly detection and OOF monitoring with automated MES-integrated alerting
• Built a new bevel-etch detection method using peak colour analysis – became permanent defence strategy

Led smart manufacturing initiatives integrating digital twins, ML models, and automation pipelines.

• Built a digital twin of TEL Celesta – ML-guided DoE optimisation reduced chemical usage by 60% and saved USD 7M+
• Saved 40+ hours/week of manual validation through simulation-based process qualification
• Qualified >10 alternate chemistries during COVID disruptions – zero production downtime across affected lines
• Co-authored 7 technical whitepapers on yield analytics, process automation, and cross-site harmonisation

Built core engineering and program delivery foundations during the Hiroshima Fab Expansion.

• Achieved USD 7.5M CAPEX savings through smart ramp planning and tool reuse across global fabs
• Developed “MAXout” onboarding framework – boosted team productivity by 20%, later scaled across sites
• Aligned Hiroshima ramp plans with Boise and Singapore through remote calibration

NUS MSc capstone – digital transformation focused on Industrial IoT and applied ML for maritime safety and enforcement.

• Detected illegal vessel mooring using geospatial analytics with GPS and tidal-corrected clustering (K-means, DBSCAN)
• Built scalable ML pipeline on AWS (S3, Lambda) for real-time anomaly detection – reduced false positives by 22%
• Delivered end-to-end prototype influencing ~$2M potential annual revenue across four ports

R&D in bio-lasing technologies – investigating biological materials as laser gain mediums.

• Engineered and tested laser prototypes integrating biological components
• Automated data collection and visualisation using MATLAB

Hydrogen evolution research under Prof. Jason Yeo – advancing catalyst science for sustainable energy.

• First to identify sulfur atoms as active sites in amorphous MoSₓ catalysts – novel experimental finding
• Executed Operando Raman Spectroscopy to examine catalytic behaviour under real-time conditions