End-to-End
Software Engineer

Problem

Business teams struggled to turn strategic goals into concrete, structured plans. The process was manual, slow, and inconsistent across teams.

Architecture

Next.js frontend with an LLM orchestration layer that structures plain-language business input into a consistent planning model, persisted on AWS.

Solution

An interactive tool where teams describe goals in plain language and AI drafts a structured plan (objectives, drivers, and initiatives) ready to refine.

Outcome

Workshops that took days of whiteboarding condensed into hours, with a consistent, shareable artifact as the output.

Problem

HR answered the same policy questions over and over in Slack, burying real casework under repetitive lookups.

Architecture

Slack events routed to AWS Lambda; documents embedded into Pinecone for retrieval; OpenAI generates grounded answers with source references.

Solution

A RAG chatbot that answers policy and benefits questions directly in Slack, citing the exact HR document each answer comes from.

Outcome

The majority of routine HR questions answered instantly without a human in the loop, with answers employees can verify against sources.

Problem

Operational knowledge about business processes lived in scattered documents and senior employees' heads.

Architecture

Amazon Bedrock hosting Claude with retrieval over curated process documentation, deployed inside the company's AWS boundary.

Solution

An internal AI chat that explains business processes, prerequisites, and edge cases, running fully within the organization's cloud perimeter.

Outcome

New employees onboard against documented processes instead of tribal knowledge, with compliance-friendly data boundaries.

Problem

Banking workloads spread across GCP and on-premise environments needed a single operational standard. Availability could not depend on where a service happened to run.

Architecture

Containerized services on GCP and on-premise Kubernetes, with centralized ELK monitoring and alerting, and Vault plus Google Assured OSS securing the supply chain.

Solution

One DevOps practice across both environments: shared pipelines, shared observability, shared security posture, delivered by one team.

Outcome

99.9% platform availability sustained across hybrid infrastructure at one of Indonesia's largest banks.

Problem

Manual deployments made releases rare, risky events. At a bank, every release also has to clear security and compliance gates.

Architecture

Jenkins pipelines with SonarQube quality gates and Nexus artifact management, deploying containers to Red Hat OpenShift with Vault-managed secrets.

Solution

End-to-end CI/CD where every service builds, tests, scans, and deploys through the same automated, auditable pipeline.

Outcome

Over 500 microservices delivered to production on OpenShift, with deployments accelerated from manual events to routine automation.

Problem

Cloud resources created by hand drifted apart across environments. Nobody could say with confidence what production actually looked like.

Architecture

Modular Terraform codebase with reusable modules per capability, remote state, and plan/apply gates wired into CI.

Solution

The entire infrastructure defined as reviewable code: every change a pull request, every environment reproducible from scratch.

Outcome

Environment drift eliminated, disaster recovery rehearsable, and new environments spun up in minutes instead of days.