Muhamad Fariz Farhan

Financial Data Team, responsible for maintaining and expanding the stock metrics and screening infrastructure used by retail and institutional investors on Stockbit.

• Agentic engineering harness:Designed and implemented a harness that autonomously delivers tickets to merged pull requests, orchestrating context gathering, implementation and review across multi-repository codebases.
• Stock data metrics:Maintained the metrics powering Stockbit's keystats, screener and financial statement modules, keeping them accurate and timely for retail and institutional investors.
• Banking sector coverage:Implemented banking-specific metrics for both conventional and sharia banks listed under the IDX index, enabling accurate screening coverage for the banking sector.
• Screener expansion:Delivered 40+ metrics on the Stockbit screener, expanding parameter-based filtering coverage for issuer discovery.

Clearing Team, working across the Compensation domain and the services under it. One of those is the Formula Calculation Service, which stores calculation formulas, fetches the data they need from multiple domains, and provides APIs to compute results dynamically.

• Compensation System:Maintained the dynamic calculation logic behind payouts, so they were computed consistently against the defined business rules.
• Global Merchant Compensation System:Architected and maintained it, tracking cumulative compensation against configurable per-merchant-group thresholds and surfacing merchants worth investigating.
• Formula Calculation Service:Architected and maintained its APIs, automating compensation and monetary adjustment, and integrated it across the Clearing subdomains so formula logic lived in one place.
• Large-scale migration:Moved database tables of 10+ billion rows to new infrastructure with zero service downtime and full data integrity.

Assigned to the Purchase Platform Tribe, responsible for maintaining and enhancing user order history, the checkout service, the cart service, and the generic transaction platform. Continuous employment through company acquisitions, GoTo in 2022 and ByteDance in 2024.

• Order History Service:Found and fixed an algorithm defect, improving response time 16× on 10% less CPU. Separately designed a multi-layer failover that held 100% uptime through datastore failures.
• Generic Transaction Platform:Was a key architect of the platform 10+ business units build on, eliminating duplicated code and cutting development time threefold, and designed 5+ generic features that moved development from code-based to config-based.
• Distributed data stores:Directed the migration of 5+ database clusters (300+ cores) across production with zero downtime, and ran the infrastructure at 15+ TB and billions of rows.
• Distributed database specialist:Debugging, planning, consultation and proofs of concept for teams across Tokopedia, including one to find a data store that could handle 300,000 RPS.
• Generic Calculation Platform:Designed it, centralising repeatable calculation logic that had been hardcoded in many places.
• Configurable rate limiting:Designed and led a rate limiter configured from request values rather than redeployed, cutting manual configuration from hours to under one.
• Cart Service:Consolidated the core module, reducing redundancy while serving millions of daily transactions at 99.9%+ uptime.
• Sharing sessions:Ran company-wide sessions on distributed databases to 300+ attendees, holding a 4.5+/5 NPS score.
• Operating system upgrade:Planned and supervised a company-wide operating system upgrade across Tokopedia's distributed databases.
• Service resiliency:Reduced unwanted binary failures to zero by introducing Go resiliency mechanisms, and strengthened monitoring and alerting to bring time-to-acknowledge under 10 minutes.

A project-based software agency whose main focus is to help clients market their products using websites such as company profiles or product catalogs.

• Team leadership:Fully managed an engineering team of more than four people, and delivered many projects for both product catalogs and company profiles.
• CI/CD adoption:Improved company product deployment productivity by introducing CI/CD tools and modernising the team's deployment practices.
• Headless CMS boilerplate:Designed and developed a headless content management system that increased delivery productivity and reduced hundreds of man-hours of duplicate effort.
• Stack modernisation:Moved the company's stack onto a newer PHP and Vue integration.

Responsible for one major project and helping across others: an itinerary sharing system built on the Google Places and Maps APIs, in Laravel and Vue.

An Information System Development Project course at the University of Indonesia, delivered for Sucofindo as the client.

• Implementation:Led the implementation phase through to deployment, and the team's programming work.
• Documentation:Wrote the documentation and tutorials for managing and maintaining the system.
• Scope:Kept implementation aligned with what was agreed at the planning, analysis and design stages.

A cooperative information system taken from design through to deployment, as the final project for the Enterprise Application Architecture and Programming course.

• Programming lead:Led the team's programming work and kept development matched to the case study.

An online class on database fundamentals, ending on how to choose between them for a given problem.

• Types and paradigmsData stores versus databases; SQL, NoSQL and NewSQL; key-value, wide-column, document, relational and graph.
• PrinciplesACID properties and the CAP theorem.
• DesignNormalisation, atomicity, indexing, and picking a paradigm that fits.

Distributed databases from first principles, then what Yugabyte does about them.

• FundamentalsDistributed scale, network partitions and the CAP theorem.
• YugabytePostgreSQL compatibility in YSQL, horizontal write scalability, all-active nodes, fault tolerance and full ACID compliance.
• APIsYSQL as fully relational SQL, YCQL as semi-relational and tuned for throughput.

Where NewSQL came from, and why "SQL or NoSQL" stopped being the question.

• ComparisonSQL versus NoSQL, and what NewSQL adds.
• DistributionHow a distributed database actually distributes.
• ResilienceReplication factor and fault tolerance.

The kinds of testing a backend team should know about, and the tools that make each cheap enough to actually do.

• CoverageTypes of testing, and where each earns its keep.
• Load testingLocust, and building in-house automation on its API.

How a Go binary fails, how to avoid it, and how to recover when it happens anyway.

• TopicsPanic and recover, fatal, and mutexes.

Concurrency in Go, and the context package that makes it manageable.

• TopicsGoroutines and using them correctly; context, and what it is actually good for.