Open to select projects

Building intelligent systems that actually ship.

I'm Riantama Putra, a software & AI engineer based in Jakarta, Indonesia. I design and build intelligent systems that ship — from LLM pipelines to the products around them.

View selected work Contact me

years shipping: 5+; years shipping
systems in prod: 20+; systems in prod
daily AI requests served: 3M+; daily AI requests served

01 / Selected Work

Systems I'm proud of

A few projects that show the range — retrieval pipelines, edge inference, event-driven backends and the tooling that keeps them honest.

Nimbus RAG

Production retrieval-augmented generation platform for enterprise document search. Hybrid retrieval, semantic caching and evaluation loops baked in from day one.

40k+ documents indexed · p95 latency 320ms

Python
FastAPI
pgvector
Next.js

Sentinel Vision

Real-time defect detection running on edge devices for a manufacturing line. Custom-trained detection models, quantized and deployed without a cloud round-trip.

45 FPS on Jetson Orin · 99.2% recall

PyTorch
TensorRT
ONNX
Go

Ledgerline

Event-driven core banking backend handling reconciliation and settlement. Exactly-once processing across services, with full replayability for audits.

12k TPS sustained · zero-loss replay

Go
Kafka
PostgreSQL
Kubernetes

EvalKit

Open-source toolkit for evaluating LLM applications: golden datasets, regression gates in CI, and drift dashboards your whole team can read.

1.2k GitHub stars · used in 30+ teams

TypeScript
Python
LLM-as-judge

02 / Experience

Where I've built

2024 — Present
Senior Software Engineer, AI Platform · Aurora Labs
Leading the team that turns model research into reliable product features.
- Built the LLM gateway serving 3M+ requests/day with cost and quality routing
- Cut inference spend 38% via semantic caching and model tiering
2022 — 2024
Machine Learning Engineer · Kanaya Tech
Owned computer-vision systems end to end — data pipelines to edge deployment.
- Shipped defect-detection models to 12 factory sites across Southeast Asia
- Reduced false-positive rate 4× with active-learning retraining loops
2020 — 2022
Full-stack Engineer · Studio Delta
Built and scaled web products for early-stage startups as engineer #3.
- Took two products from prototype to 100k monthly active users
- Introduced typed API contracts that halved integration bugs
2016 — 2020
B.Sc. Computer Science · Universitas Indonesia
Focus on machine learning and distributed systems. Graduated cum laude.

03 / About

Engineer first, then the models

I'm a software engineer who went deep on machine learning — not the other way around. That order matters: I care as much about the deploy pipeline, the p95 latency and the on-call story as I do about model quality.

For the last five years I've been building AI-powered products across fintech, manufacturing and developer tools — usually owning the path from ambiguous idea to something running in production. I like small teams, sharp problems, and systems that are boring to operate.

Ship the whole system

A model is 10% of the work. The product, data loops and guardrails around it are the job.

Measure before magic

Evals, tracing and honest baselines before reaching for a bigger model.

Fast is a feature

Latency budgets are product decisions. I design for them from the first commit.

Toolbox

Languages

Python
TypeScript
Go
SQL
Rust

AI / ML

PyTorch
LLM fine-tuning
RAG systems
vLLM
LangGraph
Evaluation & observability

Web & APIs

React / Next.js
Node.js
FastAPI
gRPC
Tailwind CSS

Infrastructure

Docker
Kubernetes
AWS / GCP
PostgreSQL
Kafka
Terraform

04 / Contact

Let's build something intelligent.

I take on a small number of projects and roles where the problem is sharp and the bar is high. If that sounds like yours, my inbox is open.

riantamaputra751@gmail.com

Jakarta, Indonesia (UTC+7)