Pipeline reliability layer
current focus 2026Designing the reliability layer for FinCense's data pipelines: state guards, idempotent retries, stale-run cleanup, pre-flight validation and metrics.
~/ backend and data engineering
Bao Ho · Hồ Gia Bảo
Bao is a Staff Engineer at Tookitaki in Singapore, designing and building backend and data systems for FinCense, the end-to-end platform for anti-money laundering (AML) and fraud prevention used by banks and other financial institutions across Asia Pacific.
At Tookitaki since 2018, he has worked on the Transaction Monitoring and Case Manager modules, delivered transaction search on StarRocks, and is now building the reliability layer for the platform's data pipelines.
He has written software since 2014, mostly in Scala since 2016: HTTP services, Spark jobs, Kafka messaging and the CI/CD around them. He favours functional Scala (Cats Effect, Doobie, Circe) because it keeps a large codebase safe to change, and mentors the engineers around him on code quality, debugging and pinning down requirements.
100M+tx/day
Transaction-monitoring benchmark, p90 end-to-end latency under 15 minutes
2025
200TPS
Sustained throughput of a real-time API he built
2023
2 min → 10 s
MySQL-backed response over 40M+ records
2018 – 2024
10 min → 30 s
Deploy time after he set up the CI/CD pipeline
2018 – 2024
The stream is a simulation of transaction monitoring, the kind of system Bao works on. The numbers under it are real.
Designing the reliability layer for FinCense's data pipelines: state guards, idempotent retries, stale-run cleanup, pre-flight validation and metrics.
Wrote the architecture and operations runbook for FinCense's copilot for AML investigators: multi-provider LLM serving, streaming, retrieval and tracing.
Built advanced transaction search on StarRocks end to end, from schema to backend APIs.
Fixed memory and latency problems in Case Manager, for example by moving pagination from the JVM into SQL.
Four migrations across compute, storage, data and services.
Added Prometheus metrics, Grafana dashboards and k6 load tests for the APIs, and built request-reply messaging over Kafka with the monitoring around it.
2018 – now
2018
2016 – 2018
2015 – 2016
2014 – 2015
2014
Most of Bao's code lives in Tookitaki's private repositories. The public ones below are smaller projects and proofs of concept, mostly around Scalatra, Kafka and testing.
docker compose up starts the whole stack.
(opens in a new tab)
2012 – 2014
Bachelor of Science in Information Technology VNUHCM University of Science, Ho Chi Minh City2008 – 2011
Associate degree in Information Technology VNUHCM University of Science, Ho Chi Minh City2016 – 2017
Functional Programming Principles in Scala Functional Program Design in Scala Big Data Analysis with Scala and Spark EPFL, on Coursera2015
Tackling the Challenges of Big Data MIT Professional Education, on edX