Case studies

Five pieces of work, described the way an engineer would describe them.

Three enterprise engagements where the client is withheld under confidentiality, and two systems built and run by EthanCorp where everything is inspectable. Context, constraints, architecture, what was actually done, and the numbers — including the ones that are only approximately good.

On confidentiality

Client names, system names and data are withheld for the three enterprise engagements. Inventing a plausible client name would make these read better and would also make them fiction. The figures shown are the ones already published on my professional profile at dattranbi.github.io.

All case studies

01 Confidential client

Enterprise data integration platform

Twenty-plus enterprise systems that each held part of the truth, connected into governed data flows with reconciliation built in — across twelve business units and five countries.

Read the case study

20+enterprise systems integrated
150+pipelines and workflows built
90%processing effort reduction

02 Confidential client

BI & decision intelligence modernisation

A five-day manual reporting cycle replaced with governed dashboards on standardised definitions — and the arguments about whose number was right replaced with one KPI dictionary.

Read the case study

5d → 0.5hreporting cycle
27+KPIs standardised
20+dashboards and reports delivered

03 Confidential client

AI workflow automation

A ten-step manual document workflow reduced to two guided steps, with an explicit human review gate — and the acceptance rate measured rather than assumed.

Read the case study

4h → ~10mtask completion time
10 → 2manual steps in the workflow
100+documents processed per day

04 Live

SAMI — retrieval-grounded iSAMS assistant

A production RAG assistant answering iSAMS questions from a curated knowledge base, with an admin console, cost visibility and a feedback loop — free and open to the iSAMS community.

Read the case study

17in-depth guides in the knowledge base
~6,400embedded vectors backing retrieval
218automated frontend and backend tests

05 In development

LangThang AI — tour advisory chatbot

A retrieval-grounded tour advisor for the Vietnamese market, built to test a two-layer NLU design: deterministic rules first, the language model only where the rules run out.

Read the case study

96backend tests (unit and integration)
2-layerNLU: rules first, LLM on ambiguity
4-stateclosing flow to a qualified lead
Dat Tran, founder of EthanCorp

Dat Tran — every one of these was designed and built by me. EthanCorp is a one-person consultancy, so there is no question about who did the work.

Recognise your own situation in one of these?

The fastest way to find out whether it transfers is a 30-minute scoping call. If it does not, you will hear that.

Response time
Within two business days
Based in
Ho Chi Minh City, Vietnam — working across Asia and remote