Skip to content

cv

I Gede Wibawa Cakramurti (Ocha)

Platform Engineer · LLM inference and MLOps

Melbourne, Australia · Australian Permanent Resident · Open to remote

download pdf

summary

Platform engineer specialising in LLM and ML inference. As Heidi Health's founding infrastructure engineer, scaled an AI medical scribe from its first AWS account to 5 data-residency regions on 3 clouds and led the security certifications hospitals and enterprises require. Now own Heidi's self-hosted speech-to-text platform, which replaced a managed vendor at lower cost, latency and error rate. Previously 4 years in Big 4 risk, audit and M&A.

selected impact

experience

Heidi Health (formerly Oscer), Melbourne

Oct 2021 - Present

AI medical scribe used in 190+ countries, ~2.8M patient visits a week (Sep 2026).

Senior Cloud Engineer, Engineering Platform

Aug 2025 - Present
  • Designed and built the self-hosted speech-to-text platform (Triton, fine-tuned Parakeet, EKS GPU clusters, cross-region PrivateLink), taking it from prototype to production in one quarter.
  • Led the vendor cutover through shadow traffic, a feature-flagged regional ramp and per-request fallback.
  • Migrated the fleet from L40S to RTX PRO 6000 (g7e) with zero downtime, and rebuilt capacity management around reserved GPUs with Karpenter, NodeOverlay and automated reservation booking.
  • Built alerting, runbooks and a progress-based liveness probe that recovers stalled Triton pods in ~14 minutes without paging; diagnosed and fixed shared-memory exhaustion and poison-input failures.
  • Decommissioned the legacy ECS platform for 3 services with zero downtime, launched Azure production regions in Japan, the US and the UAE, and replaced broad AWS PowerUser access with 4 least-privilege PIM roles.

Acting Engineering Manager, DevOps

Feb 2025 - Jul 2025
  • Managed 3 engineers (junior, mid, senior); introduced Linear request intake, a ticketed help desk and a 4-round DevOps interview loop.
  • Built the MongoDB Atlas platform and migrated terabyte-scale data off DocumentDB in 5 regions using DMS change data capture and a rollback path; cut search p95 latency ~80% and enabled vector search.

DevOps / Machine Learning Engineer, founding infrastructure engineer

Jan 2022 - Jan 2025
  • Owned infrastructure, CI/CD, databases, observability, security, compliance and IT as the sole infrastructure engineer, from the first AWS account through the Scribe launch (Oct 2023).
  • Launched 4 data-residency regions in 5 months (UK, Canada, US, EU; 2024) and migrated existing users and data to their home regions.
  • Led ISO 27001 (2023; 2022 transition 2024; 2025 audit), SOC 2 Type II (2024) and Cyber Essentials (2024) certification, largely single-handed; now supervise renewals.
  • Established BCP and DR, asset tagging, incident response and customer security questionnaire processes.
  • Designed the first AWS platform in Terraform, CI/CD for all backend services, a Go API Gateway authoriser and Datadog observability; ran company IT (laptops, MDM, network, DNS).

Machine Learning Engineer (intern to full time), Oscer

Oct 2021 - Jan 2022
  • Cut the patient/bot chat data, training and deployment cycle from ~50 to ~4 hours; built model CI/CD on MLflow.

Interviewed backend system-design candidates and most platform team hires.

Earlier career

  • Cultural Infusion, Melbourne: Junior Data Scientist · Nov 2020 - Apr 2021. Built a privacy risk assessment framework for a cultural diversity platform.
  • PwC, Jakarta: Senior Associate, Risk Assurance Data & Analytics · Jan 2019 - Jan 2020. Automated a bank's country-risk reporting, cutting preparation time 80%; led a team of 2-4 with PwC Japan.
  • EY, Jakarta: Consultant, Advisory, and Junior Auditor · Sep 2015 - Dec 2018. Implemented a risk management and internal audit framework for Bank Indonesia; trained 25 staff up to Deputy Director level.

talks

skills

Inference and MLOps
NVIDIA Triton, NeMo ASR, GPU Operator, MIG, DCGM, KEDA, Karpenter, Gateway API, MLflow
Cloud and platform
AWS (EKS, Organizations, PrivateLink, IAM, DMS, Lambda), Azure (AKS, Entra ID), GCP (Vertex AI), Terraform, Helm, GitHub Actions, MongoDB Atlas
Reliability and security
Datadog, OpenTelemetry, Prometheus, incident command, BCP and DR, ISO 27001, SOC 2, Cyber Essentials, Vanta
Languages
Python, Go, Bash, SQL

education