Dawei Ma

Senior Full-Stack / Product Engineer — AI-native Products, Cloud & Architecture

Xi’an, China · Open to global remote roles · AI application engineering / Agent workflows / senior full-stack & platform engineering

github.com/madawei2699 · Blog · myInvestPilot · [email protected]

Summary

I am a senior full-stack engineer and product-minded architect with 10+ years of experience building and operating software systems for enterprise clients and independent products. At Thoughtworks, I have worked remotely with overseas clients on real-estate, advertising, data, messaging and infrastructure systems, including event-driven notification redesign, cross-team migrations, observability, security scanning and production issue diagnosis.

In parallel, I build AI-native products as an independent builder. My current focus is LLM applications, Agent workflows, tool/API design, schema-constrained execution, AI-assisted development and the engineering boundary between prototypes and long-running production systems. I am interested in remote roles where I can combine product judgment, system design and hands-on engineering: AI application/product engineer, Agent workflow builder, senior full-stack engineer, or platform engineer.

Scale hints: enterprise systems with multi-service architecture and zero-downtime releases; independent AI SaaS products with real users, paying customers, automated workflows and LLM-powered features.

Skills

Backend/Platform: TypeScript/Node, Python, Java/Kotlin, Clojure, Elixir, Go; Cloudflare Workers/Queues/R2/D1/Durable Objects, AWS, GCP, Docker, CI/CD
Frontend/Product: React, Next.js, Tailwind, WebRTC integration, product prototyping, UX-minded engineering
AI/Product Engineering: LLM applications, Agent workflows, tool/API design for LLMs, JSON Schema, domain DSLs, prompt/tool documentation, AI-assisted development
Data/Automation: PostgreSQL/Supabase, Redis, BigQuery, event pipelines, backtesting workflows, email automation
Engineering Practices: System architecture, ADRs, Mermaid architecture diagrams, incident response, cross-team collaboration, observability, security scanning

Experience

Senior Consultant — Thoughtworks (May 2019–Present)

Full-stack engineering / architecture / technical consulting for overseas enterprise clients.

Large Real-Estate Marketplace in Australia (Remote, Nov 2022–Present)
Stack: TypeScript/Node, React, BigQuery, GCP, Braze, Docker, CI/CD, fp-ts
Domain: B2B advertising products, agency systems, listing management, event pipelines and notification workflows

Built and maintained ad platform features — customer ad purchase interfaces, ad serving, impression tracking and reporting pipelines using BigQuery. Integrated with event-driven infrastructure and collaborated across teams to deliver business features.
Redesigned a notification system — replaced a legacy MQ-based solution with a more flexible event-driven architecture and coordinated multi-team zero-downtime migration, improving maintainability and future feature delivery.
Solved production blockers — diagnosed infrastructure and deployment issues involving VPC peering, IAM credential propagation and failed notification recovery; wrote emergency scripts and technical deep-dives to unblock delivery.
Improved engineering communication — introduced Mermaid architecture diagrams in ADRs, documented complex data flows and helped establish patterns for distributed messaging systems.

German Automotive OEM (Remote, Jan 2021–Sep 2022)
Stack: Clojure (nginx-clojure), Kotlin/Java, AWS, AppDynamics, Nginx, GoCD, Phrase

Led investigation of random 504 errors by testing across application and AWS infrastructure layers and coordinating with infrastructure teams and open-source maintainers.
Enabled BlackDuck security scanning for Clojure projects despite unsupported tooling; the solution was adopted across teams.
Corrected inaccurate monitoring for Clojure services and improved performance through web server tuning.
Contributed a bug fix to Google’s Kaniko project, which was approved and merged upstream.

Leading China ICT Company (Apr 2020–Dec 2020)

Maintained and supported enterprise backend platforms and logging SDK integration using Java/Spring Boot; worked directly with multiple integration teams on compatibility, troubleshooting and documentation.

Major Airport Advertising Company (Aug 2019–Oct 2019)

Acted as Tech Lead and guided the team through backend microservices refactoring.

Senior Java Developer — Yeahmobi (Apr 2018–May 2019)
Stack: Java/Spring Boot, MySQL, Redis, RabbitMQ, AWS, Python

Built backend platforms for ad campaign management — integrated third-party advertising APIs for campaign creation, reporting, data synchronization and workflow automation.
Developed data pipelines — built web crawlers and collaborated with the ML team for data classification and downstream content distribution.
Optimized database and caching — improved query and caching strategies, reducing response time by around 50%.

iOS Developer — Fengyu (Apr 2016–Apr 2018)

Founder — Financial Services Startup (2015–2016)

Selected Products & Open-source Projects

myInvestPilot / Chat2Invest — AI-native investment research and strategy-learning product
www.myinvestpilot.com · www.chat2invest.com · AI architecture notes

Independent SaaS product for retail investors to learn, research and review strategies; it does not provide investment advice.
Served 3k+ users, nearly 200 paying users and 450k+ cumulative page views; 122k+ page views in 2026.
Product capabilities include official simulated portfolios, custom simulated portfolios, strategy primitives, historical backtesting, market analysis, risk metrics and automated email alerts; sends nearly 1k strategy/market emails per day.
Designed a structured JSON DSL for strategy primitives, allowing LLMs to generate, explain and diagnose strategy configurations while deterministic engines handle schema validation, backtesting, metric calculation and result reproducibility.
Built local Agent experiments around the strategy engine: LLM-readable quickstart docs, JSON Schema, examples, skills and local workflows for generating configurations, running batch backtests, analyzing results and iterating parameters.
Runs as 20+ repositories/services across strategy engine, backtesting, market data, AI analysis, notifications, user systems and infrastructure; online AI features consume about 15M tokens/month with low LLM operating cost.

LLMAgent — Experimental domain Agent abstraction library
github.com/i365dev/llm_agent

Elixir-based Agent abstraction experiment focused on message processing, tool calling, long-running tasks, context management, provider plugins and dynamic workflow orchestration.

myGPTReader — Open-source AI reading assistant
github.com/myreader-io/myGPTReader

Early open-source AI application with 4k+ GitHub stars. It reads and summarizes webpages, documents and YouTube videos in Slack, and supports voice conversation, prompt templates and news summaries.

free4.chat — Minimal real-time audio chat product
www.free4.chat · github.com/i365dev/free4chat

Open-source no-signup real-time voice chat app with 1.1k+ GitHub stars. Explored multiple architecture versions including Go + Pion SFU, Elixir + Membrane and a Cloudflare-based stack; continuing to explore room-level AI conversation scenarios.

Writing & Talks

Maintain BMPI.dev, a long-running personal blog since 2019 with 1M+ cumulative page views, covering software architecture, cloud engineering, AI-native products, investing and independent building.
Former AWS Community Builder; published articles on Kubernetes, Serverless, microservices and cloud architecture.
Gave public talks and internal sharing sessions on ChatGPT application development, AIGC learning, WebRTC, software engineering practices and technical influence.

Education

B.E., Computer Science & Technology — North China Electric Power University (2008–2012)

Languages

English: Professional Working Proficiency
Mandarin: Native