reduction in feature lead times across the GCP ML platform
Montu — GCP ML Platform
methodology
DORA metrics, pre/post Kubeflow + GitOps migration
~/runtime
Career runtime · 7 roles
Industry × Role × Quantified outcomes. Most-recent first.
-
Montu
Staff Machine Learning Engineer · Privacy-by-Design Architect
Healthcare 2023 – Present67% 97%+ reduction in change-failure rate after platform consolidation
Montu — DORA Outcomes
methodology
Incident DB diff, 6-month rolling
93% clinical NLP accuracy (outperforming Google Healthcare NLP by 14 pts)
Montu — Clinical Document Intelligence
methodology
Held-out clinician-labelled validation set
0.87+ F1 PII redaction score on clinical log sanitisation
Montu — Privacy-by-Design Logging
methodology
Held-out clinician-annotated PII corpus, micro-F1
Tech
- GCP (GKE, VertexAI, Pub/Sub)
- Kubeflow + TFx + DVC pipelines
- PyTorch · vLLM · FlashAttention
- GraphRAG + LangGraph + DSPy
- Postgres · Snowflake · Redis · Kafka
Leadership
- Hired & mentored Data Scientists and ML Engineers across multiple pods
- Authored RFCs for clinical NLP architecture and Privacy-by-Design logging
- Cross-functional Build vs Buy negotiation with Product/Compliance/Legal
-
Amber Electric
Senior Machine Learning Engineer · FinOps Strategist
Energy 2021 – 202340% average infrastructure cost optimisation across 10k+ energy sites
Amber — Energy Forecasting Platform
methodology
GCP billing diff, 6-month rolling, post-FinOps refactor
Tech
- GCP (GKE, Pub/Sub, BigQuery, Dataflow)
- Time-series forecasting (ARIMA · Prophet · DeepAR)
- Event-driven architecture · Kafka
- Terraform · Helm · Datadog
Leadership
- Led event-driven architecture transition for grid-stability forecasting
- Authored FinOps strategy reducing cloud spend across 10k+ energy sites
-
Linktree
Machine Learning Engineer — Recommender Systems
Consumer Tech 2020 – 202133% link adoption gains via the personalised link-recommender
Linktree — Two-Tower Recommender
methodology
A/B test, 30-day window, user-cohort holdout
18-23% profile subscription uplift across recommender experiments
Linktree — Profile Subscriptions
methodology
A/B uplift, 90-day cohort
Tech
- Two-Tower Architecture · Collaborative Filtering
- Python · TensorFlow Recommenders
- AWS (SageMaker, Lambda, DynamoDB)
- Airflow · Feast feature store
Leadership
- End-to-end recommender ownership from offline metrics to online A/B
- Cross-functional product negotiation on metric framing and guardrails
-
ANZ Bank
ML Engineer — Enterprise ID Fraud Detection Platform
Finance 2019 – 2020Production enterprise identity fraud detection on Kubeflow / GKE
ANZ — Enterprise ML Platform
methodology
Production deploy across multiple business units
Tech
- Kubeflow Pipelines · GKE · Tekton
- XGBoost / LightGBM ensembles
- Compliance-grade RBAC + audit pipelines
Leadership
- Enterprise ML platform architecture across regulated business units
-
nib Group
ML Engineer — Health Insurance Platform
Healthcare 2018 – 2019Platform SageMaker platform extensions powering claims ML workloads
nib — ML Infrastructure
Tech
- AWS SageMaker · Lambda · S3
- Clinical claims data engineering
- Schema validation · drift monitoring
-
HammondCare
ML Engineer — Customer Triaging
Healthcare 2017 – 2018Live customer-request triaging chatbot in clinical operations
HammondCare — Triaging Chatbot
Tech
- NLU pipelines
- Customer-request classification
- Production chatbot integration
-
Eliiza
ML Engineer · ML Guild Mentor
Consulting 2016 – 20172nd place at Cricket Australia DataJam 2020 (cross-functional team lead)
Eliiza — Cricket Australia DataJam
Framework "Thea" document mining framework adopted across consulting engagements
Eliiza — Thea Document Mining
Tech
- Document mining (Thea framework)
- NLP pipelines for unstructured legal/clinical text
- Cross-functional consulting delivery
Leadership
- ML Guild mentoring across the consulting practice