Senior Full Stack Engineer — LLM & AI Systems @ Novo AI
Building AI-integrated systems that automate insurance claims, pre-authorization, and retrospective case reviews. Previously Technical Lead at MarinaChain, full-stack/mobile/AI at KinderCastle, backend at Live Play Mobile, and full-stack at Pintarnya.
Top Graduate of Institut Teknologi Bandung (3.99 / 4.00 CGPA), Informatics '20.
- AI & LLM systems — medical OCR pipelines that understand context (not just read text), agentic workflows aware of ICD-10, ICD-10-PCS, HCPCS, DRG, TOSP, Aetna CPB, and EviCore Cigna guidelines, RAG over vector DBs, and AI-first error triaging wired to Sentry + traces.
- Full-stack products — React / Next.js / Vue on the frontend; Node, Go, Python (FastAPI, Django, Express) on the backend; PostgreSQL, MongoDB, Redis, DynamoDB, Elasticsearch, Milvus for storage.
- Mobile apps — Flutter (Bloc) and React Native, with clean-architecture patterns across data sources, repositories, use cases, and widgets.
- Cloud & DevOps — AWS (EC2, RDS, ElastiCache, Lambda, S3, CloudFront, Amplify), GCP (GKE, Cloud Run), DigitalOcean, Cloudflare, Docker, Kubernetes, CI/CD with GitHub Actions, OpenObserve for observability.
Novo AI — Senior Full Stack Engineer (LLM & AI Systems)
- Researched 10+ OCR models for handwritten medical documents; shipped a context-aware approach that lifted test pass rate from 65% → 90%+.
- Built a multi-agent pipeline for insurance claim / pre-auth / retrospective review, handling 1000+ documents daily (including 50+ page invoices) with medical-guideline awareness.
- Improved line-item extraction across long invoices — reliable tax/discount distribution without double entries — from 70% → 92.5% on tests.
- Shipped an AI-first triage system wiring Sentry, traces, logs, and customer feedback into agent skills, making bug-ticket creation ~10× faster.
MarinaChain — Technical Lead (BE/FE/ML/DevOps)
- Led a cross-functional team of 4 (FE, BE, ML), ran 1:1s, ran hiring loops, scoped sprint work.
- Migrated all microservices to AWS (EC2, RDS, ElastiCache, S3, CloudFront, Lambda).
- Cut monthly AWS billing by ~40% via right-sizing + EC2/RDS savings plans.
- Stood up centralized observability across 7+ services with OpenObserve.
- Shipped CI/CD for every microservice via GitHub Actions + Docker — ~10× faster deploys vs manual SSH.
- Migrated Vue 3 CLI + Babel → Vite: 30% smaller bundle, 80% faster builds, fixed 232 ESLint errors along the way.
- Built MarinaDocs frontend — 6+ pages, DnD filesystem, streaming chatbot events, PDF preview, full Korean/English i18n, headless a11y-first components.
Live Play Mobile — Backend Engineer
- Sole backend engineer on a live bingo game with complex side-bets, third-party casino integrations, serving ~1k+ concurrent users.
- Wrote the math models + Monte Carlo simulations — >1M rounds simulated in under 10s — to prove RTP matches casino targets.
- Shipped game quests generating $15k+/day and $12k+/day in in-game revenue.
- Cut DynamoDB monthly cost by ~30% by tuning RCU/WCU.
- Made Serverless deploys ~2× faster via shared Lambda layers and caching.
KinderCastle Indonesia — Middle Software Engineer (Full-stack, AI, Mobile)
- Built the Xendit-integrated subscription system (late fees, overtime, leaves, churn) — replaced manual tracking, ~20× operational speedup across 500+ children daily.
- Shipped automated daily revenue recognition from child attendances — 500+ journal entries/day, zero manual input.
- Built an AI CV parser (vision LLMs) — HR went from weeks → minutes on 1000+ CVs.
- Built an internal RAG + multimodal chatbot for daily staff use.
Pintarnya — Software Engineer (Fullstack)
- Designed Pintarnya AI CV Review — LLM + OCR + multimodal — serving ~10k CV reviews/month at low per-CV cost.
- Migrated frontend from Cloud Run → GKE, cutting cost by 2×+.
- Solved race conditions in money-handling services (Cari Cuan, Cari Promo) and integrated affiliate schemes with Dynata and Toluna.
- Cut DB load ~8× via Redis caching + bulk updates.
- Earlier, tuned Pintarnya's job-search query: DB CPU 90% → <50% — bridge fix before the Elasticsearch migration.
Languages · TypeScript · Go · Python · Dart · Ruby · Haskell
Frontend · React · Next.js · Vue 3 · Pinia · TanStack Query · Tailwind · Storybook · Cypress
Backend · Node (Express) · Go · FastAPI · Django / DRF · SQLAlchemy · Sequelize · Celery
Mobile · Flutter (Bloc) · React Native
AI / ML · OpenAI API · LangChain · ReAct agents · RAG · Milvus · OCR pipelines · multimodal LLMs
Data · PostgreSQL · MySQL · MongoDB · DynamoDB · Redis · Elasticsearch
Cloud / DevOps · AWS · GCP (GKE, Cloud Run) · DigitalOcean · Cloudflare · Docker · Kubernetes · GitHub Actions · Caddy · Nginx · OpenObserve
- Sugarsmith.co — Shopify + Resend storefront for a Singapore client selling Sanrio macarons.
- Daily Wins — React + Express + Postgres app for tracking daily wins and tomorrow's goals, JWT auth.
- KataBaku — vanilla JS study tool for Indonesian UTBK pronunciation prep.
- Psychomotor Vigilance Task Test — Next.js + Google Sheets API for running and storing PVT tests.
- Samuel Cahyawijaya's personal site — customized al-folio Jekyll template.
- LinkedIn — marchotridyo
- GitHub — @acomarcho



