Technical Delivery Lead | Software Engineer

I build reliable systems and practical AI agent workflows, while helping teams ship with less chaos.

I'm David Bong. Most of my recent work has focused on AI agent system development for B2B automation, alongside backend integration and technical delivery. I focus on multi-agent orchestration with Strands, reliable execution harnesses, and practical eval workflows so agent behavior is safer and easier to measure.

Based in

Penang/Sarawak, Malaysia

Enjoy working on

Systems with real operational impact

Lately

AI agent systems, integrations, and delivery

300k+ daily transactionsMulti-agent orchestration for B2B workflowsExecution harnesses and eval workflows for agents
View Featured WorkContact David
Professional Summary

The short version of how I work.

Over the last two years, I've worked across backend integration, Flutter delivery leadership, QA execution, and AI agent system development. Recently, much of my focus has been B2B automation with multi-agent orchestration, reliability harnesses, and early eval workflows. I'm most energized by work that combines strong technical foundations with practical operational thinking.

Where I create value

I build systems that help businesses move information reliably, especially where accounting, ERP, and operational workflows meet. I also build AI agent workflows that are designed to execute reliably in production, and I enjoy creating delivery habits that improve release visibility, reduce churn, and help teams work with more confidence.

I like owning both the technical work and the shape of the delivery around it.

I help teams adopt agent best practices that are practical enough to use every week.

I care about explaining things clearly, not just building them well.

I want the people I work with to feel more confident, not more overwhelmed.

Featured Work

A few projects that say a lot about the kind of work I enjoy.

The recent focus is AI agent systems for B2B automation, alongside integration and delivery work. Each project gives a quick sense of what I built, what I owned, and what made it meaningful.

Professional WorkRecent focus

AI Agent Systems for B2B Automation

Building Strands-based multi-agent workflows for order and document operations in B2B environments.

I worked on a Python/FastAPI agent service built on Strands, with a master orchestrator routing requests across specialist agents for cart, customer, product, pricing, logistics, and document flows. The focus was reliable execution in production settings: mode-aware orchestration, steering guardrails, structured outputs, and evaluation harnesses that make agent behavior easier to verify.

My role: Software Engineer / AI Agent Developer

Impact: Helped move AI automation from ad hoc prompting toward a clearer engineering system with predictable orchestration, safer tool use, and measurable quality checks.

Platform

Python + FastAPI

Orchestration

Multi-agent

Reliability

Steering + eval harnesses

Highlights

Mode-aware orchestration

I worked on orchestrator flows that adapt behavior by context and route tasks to the right specialist agent.

Steering guardrails

I implemented and refined steering rules to block avoidable tool mistakes and enforce safer response behavior.

Evaluation foundation

I helped set up eval harnesses and golden-case workflows so quality discussions stayed evidence-based.

Team best-practice enablement

I contributed documentation and working standards that made agent implementation patterns easier for the team to reuse.

Outcomes

  • Delivered orchestration flows that route work to focused agents instead of overloading one agent with every task.
  • Introduced steering handlers to reduce redundant or invalid tool calls and enforce key verification steps before final replies.
  • Set up practical eval workflows using golden fixtures and DeepEval-style checks, alongside unit and harness tests.
  • Improved team consistency through architecture docs, tool docstring standards, and reusable development patterns.

There's more detail inside if you feel like exploring.

Professional WorkJun 2025 - Present

Product Delivery Leadership

Helping a Flutter team ship faster with clearer guardrails, better communication, and stronger delivery discipline.

When I joined the Flutter team, I focused less on chasing isolated fixes and more on improving how the team worked together day to day. That meant writing practical guardrails, encouraging more proactive communication, improving QA-dev alignment, and becoming a clearer bridge between customer support and engineering.

My role: Scrum Master / Software Engineer

Impact: Improved release visibility, made delivery feel more disciplined, reduced avoidable confusion between teams, and helped the team move faster with fewer ad hoc problems.

Focus

Faster, clearer delivery

Main lever

Playbooks and habits

Cross-team bridge

CS, QA, and dev

Highlights

Playbook thinking

I wrote guardrails that made the team's delivery expectations more visible and repeatable instead of depending on memory or urgency.

Coding consistency

I wrote playbooks to encourage more consistent coding patterns, which made debugging easier and delivery work less messy across the team.

Proactive communication

I pushed for earlier conversations around risk, blockers, and quality so fewer issues appeared at the last moment.

Cross-team alignment

I helped QA, developers, and customer support share context more clearly so delivery decisions were less fragmented.

Calmer support-engineering flow

I replied quickly to customer support, helped de-escalate tension, and pushed conversations toward clearer context instead of urgency alone.

Outcomes

  • Gave the team more visibility into what was shipping and where work was getting stuck.
  • Helped developers and QA communicate earlier so issues were surfaced before they became release noise.
  • Reduced some of the friction between customer support and engineering by making expectations and feedback clearer.
  • Helped customer support get faster replies and calmer conversations when issues felt urgent.
  • Improved debugging and day-to-day delivery consistency by introducing clearer coding patterns through written playbooks.

There's more detail inside if you feel like exploring.

Professional WorkJun 2025 - Mar 2026

Boostorder Rep V3

An offline-first Flutter app built to help sales teams keep working smoothly even when connectivity is unreliable.

This product was built for B2B sales workflows where people are constantly moving, switching contexts, and dealing with imperfect internet. A big part of the work was making the app feel smooth and dependable, so users could still capture work, review information, and move through their day without the product fighting them.

My role: Software Engineer / Flutter Contributor

Impact: Improved the reliability and usability of a sales automation app by supporting offline-first behavior, smoother UX, and a more practical mobile workflow for field users.

Platform

Flutter mobile app

Experience

Offline-first workflow

Focus

Smooth B2B sales UX

Highlights

Offline-first thinking

I helped shape a product experience that stayed useful even when connectivity was inconsistent.

Smoother field workflow

The app was improved around the pace of real sales work, with less friction in common mobile actions.

Practical UX focus

I cared about making the product feel dependable and pleasant to use, not just technically correct.

Technical workaround thinking

I solved racing issues and found practical workarounds for technical limitations so the app could stay stable without sacrificing the user experience.

Outcomes

  • Made the app more dependable in real working conditions where internet access could be weak or inconsistent.
  • Helped shape a smoother user experience so common actions felt faster and less frustrating.
  • Supported a product experience that matched the pace and reality of B2B sales work instead of assuming perfect connectivity.
  • Reduced edge-case instability by solving racing issues and creating workable paths around technical limitations in the stack.

There's more detail inside if you feel like exploring.

Professional WorkJan 2024 - Jun 2025

Boostorder Cloud Sync

A RabbitMQ-driven integration platform built for near real-time accounting and ERP synchronization.

Cloud Sync is the backbone that moves transactional data between Boostorder Cloud and external accounting systems with reliability, observability, and support workflows in mind.

My role: Full Stack Software Engineer / Scrum Master

Impact: Supports 300k+ daily transactions and serves enterprise customers with high operational expectations, including Malaysian e-Invoicing workflows.

Daily volume

300k+ transactions

Integration style

Near real-time sync

Core infrastructure

RabbitMQ-driven

Highlights

Architecture ownership

Co-architected the platform and stayed close to implementation details instead of treating design and delivery as separate jobs.

Delivery accountability

Owned delivery across design, code review, documentation, and production support.

Enterprise usefulness

Built with the operational realities of accounting and invoicing workflows in mind.

Outcomes

  • Improved confidence in production sync workflows with better failure visibility and payload resync support.
  • Helped enterprise clients adopt accounting integrations without relying on fragile manual handoffs.
  • Owned delivery across system design, implementation, code review, documentation, and support.

There's more detail inside if you feel like exploring.

Professional WorkJan 2025 - Jun 2025

Dynamic Mapping Engine

A no-code configurable mapping engine for transforming data across APIs, databases, and files.

Dynamic Mapping gives teams a configurable way to reshape and route data without rewriting integration logic for every new format or client need.

My role: Software Engineer

Impact: Reduced the need for one-off mapping code by introducing a reusable rules-driven transformation approach.

Experience

No-code configuration

Input types

API, DB, Excel

Core idea

Rules-driven transformation

Highlights

Reusable system design

Turned repeated transformation work into a reusable capability instead of a string of custom patches.

Integration leverage

Designed around the kinds of inputs teams actually deal with in business systems: APIs, databases, and files.

Maintainability

Improved long-term delivery speed by reducing how much logic had to stay hardcoded.

Outcomes

  • Created a more scalable path for integration customization.
  • Improved maintainability by moving mapping behavior out of hardcoded per-client logic.
  • Made future extension easier by defining a flexible transformation layer.

There's more detail inside if you feel like exploring.

Personal Project2024

Elysium NFT Marketplace

A full-stack blockchain marketplace simulation for minting and trading NFTs.

Elysium was a hands-on exploration of smart contracts, decentralized application architecture, and web performance in a user-facing marketplace experience.

My role: Full Stack Developer

Impact: Built to deepen full-stack and blockchain engineering skills across frontend, backend, smart contracts, and testnet deployment.

Stack

Vue, Express, MongoDB

Blockchain

Solidity + Hardhat

Optimization

IndexedDB caching

Highlights

Cross-stack build

Designed and implemented the app across frontend, backend, database, and blockchain layers.

Performance thinking

Used IndexedDB caching to improve metadata loading and reduce repeated fetch overhead.

Exploratory depth

Used the project to build practical confidence with smart contracts and decentralized app architecture.

Outcomes

  • Strengthened practical understanding of smart contract workflows and dApp architecture.
  • Explored frontend and backend integration patterns in a decentralized context.
  • Applied caching strategies to improve metadata loading performance.

There's more detail inside if you feel like exploring.

Leadership and Execution

Improving both the systems and the way teams ship them.

My strongest work sits at the intersection of engineering and execution. I enjoy building dependable systems, but I also care about how teams gain visibility, reduce delivery friction, and create more predictable outcomes.

Technical delivery leadership

I stay close to architecture and implementation while also creating clarity around scope, execution, and production support.

Flutter team leadership

As Scrum Master, I introduced a discipline system that improved release visibility, reduced ad hoc issues, and helped the team deliver with more confidence.

QA process improvement

I took on Scrum Master responsibilities for QA, introduced more data-driven quality tracking, and worked to reduce test-case flakiness.

Team-wide operating playbooks

I created a Scrum Master playbook to help managers and team leads apply more consistent delivery practices across teams.

AI agent team enablement

I help the team adopt practical best practices for agent orchestration, harness design, and evaluation so experiments become reliable delivery habits.

Books That Shape How I Work

The ideas I return to when I think about execution, growth, and collaboration.

I keep this section because the best engineering work is shaped by how we think, not just the tools we use. These books reflect the mindset I try to bring into delivery, ownership, and teamwork.

Reading Notes

The 7 Habits of Highly Effective People

by Stephen R. Covey

This book shapes how I think about ownership, discipline, and relationships, especially when work gets ambiguous or fast-moving.

Key themes

  • Be proactive and take ownership instead of waiting for cleaner conditions.
  • Start with the end in mind when designing systems, teams, or personal goals.
  • Seek first to understand, then to be understood when working across teams.

Reading Notes

The Phoenix Project

by Gene Kim, Kevin Behr, and George Spafford

This book gave me a more practical way to think about flow, bottlenecks, and why healthy delivery systems matter just as much as technical ability.

Key themes

  • Local optimization can still hurt the whole system when the real bottleneck stays untouched.
  • Good work flows better when teams reduce handoff friction and make problems visible early.
  • Operations, development, and business priorities work best when people treat delivery as one shared system.
Let's Connect

Happy to chat if you want to build something interesting together.

I'm especially drawn to work where technical depth, thoughtful execution, and good collaboration all matter at the same time.

EmailGitHubLinkedIn