Senior Data Scientist · Jakarta, ID
Chandra
Liuswanto
Problem solver. I turn messy, high-stakes problems into systems that ship.
Three years building production AI and data systems across financial services and e-commerce. I care about the result on the other side of the model: time saved, risk removed, revenue moved.
~3 hrs → under 60s
every month, in production
e-commerce recommendations
faster live risk decisions
Selected Work
Personal builds. Two lean on AI; one is straight systems engineering — same instinct, different tools.
Demo — coming soon
01
Agentic AI
Full-stack
Private
Deckwright
A prompt becomes a finished slide deck.
A chat-style app that interviews you about your topic, then runs a
writer/critic agent loop to draft slides and export them to
.pptx. The hard parts are the engineering around the
model: durable background jobs that survive a page reload, live
progress over SSE, a sandboxed renderer, and a deterministic spec
linter that catches layout overflow and feeds the exact math back to
the agent to fix.
- React + Vite
- Express 5
- Postgres
- OpenAI-compatible LLM
- JWT / LDAP auth
Demo — coming soon
02
Agentic AI
Distributed systems
Private
auto-ocr
Document extraction that teaches itself.
A self-hosted OCR platform with human-in-the-loop correction and a meta-agent that writes, tests, and versions its own extraction code per project. Once operators correct enough samples, the agent generates a custom runner and a configurable scoring pipeline decides which mismatches actually count.
- FastAPI
- Celery
- Redis
- Postgres
- React + TS
- Docker
Demo — coming soon
03
Full-stack
Fintech
Private
Bank Statement Analyzer
Structured financial data from raw statements.
A FastAPI + React tool that parses Indonesian bank statements into clean, analyzable financial data — built for a domain where formats are inconsistent and accuracy is non-negotiable. No model required; just careful parsing, validation, and UX.
- FastAPI
- Python
- React
- JavaScript
Repositories are private — happy to walk through any of them on request.
Experience
Where the work met real stakes — production systems, real money, real risk.
Data Scientist · Senior Assistant Manager
OCBC Indonesia
- LLM document pipeline — cut processing from ~3 hrs to under 60s (98%↓) at 90%+ accuracy, across 9,000+ statements a month. Coordinated directly with Credit Risk and Relationship Management.
- Financial OCR → Straight-Through Processing — 100+ docs/day at 98% end-to-end success and 92% per-field accuracy, with automated prompt generation and quality controls.
- On-prem GPU infrastructure — architected the LLM serving backbone for multiple real-time production apps, with lower latency and tighter data security.
- AML risk layer — a secondary anti-money-laundering detector that suppressed false positives at 85% accuracy, streamlining live risk decisions.
Data Scientist
Otten Coffee
- Recommendations — consumer-behaviour ML driving a 2% daily sales uplift and a 300% increase in add-to-cart conversion.
- Retention — segmented behaviour analysis lifting e-commerce retention 25% through targeted personalization.
- Data platform — modernized the stack (internal/external DBs, scraping, APIs) for a 60% gain in analysis efficiency, plus pipeline design, monitoring, BI, and orchestration on the cloud.
M.Sc. Aerospace Engineering · Cum Laude · GPA 3.75/4.00
Bandung Institute of Technology
Heavy on modelling, simulation, and the kind of first-principles problem solving I still lean on when a system has to be reasoned about, not just trained.
About
Senior Data Scientist with 3+ years delivering production AI across financial services and e-commerce. I’m a pragmatic problem solver — I translate tangled technical challenges into scalable systems that drive efficiency, reduce risk, and grow the business. AI is one tool in that work, not the point of it.
Leadership & Strategy
- AI Governance
- Project Management
- Stakeholder Management
- Tech Stack Architecture
- Red Teaming
Data Platforms & Storage
- PostgreSQL
- BigQuery
- MySQL
- Redis
- S3
- Hive
Pipelines & Orchestration
- dbt
- Airbyte
- Prefect
Analytics & Monitoring
- Apache Superset
- Metabase
- Looker Studio
- Grafana
- Prometheus
Infrastructure & MLOps
- Docker
- Kubernetes
- GPU Inference
- API Design & Deployment
Interests
- LLM Governance
- AI Red Teaming
- Coffee
Get in touch
Open to opportunities and collaborations — especially where the problem is hard and the impact is measurable.