Founding engineer/tech lead of several teams in Amazon Halo tackling complex problems across the full stack — Science, UX, Connectivity and Firmware — to deliver value for customers.

Tech stack: AWS, Android/Kotlin, iOS/Swift, FreeRTOS/C, React Native/TypeScript, Rust, Python.

• Designed and led a team of engineers to build services powering the next generation of Halo products.
• Designed and led a team of engineers to build an ML accuracy evaluation framework adopted by at least 5 teams across the Halo org (i.e., Biometrics Science, Edge Compute, Wearable Firmware, Firmware Automation, and Cloud Compute).
• Designed and led a team of engineers to build new-to-the-world features spanning the Halo fitness wearable, the companion app and the backend services.
• Wrote critical-path Rust code for Halo's edge compute systems: 1) low-latency ML workflow engine for real-time emotion inference of audio samples and 2) continuous improvement of Alexa Speech ML models running on mobile devices.
• Proposed and standardized 1) the feature gating process and 2) the frontend coding practices across the org of 100+ engineers.
• Actively mentoring and guiding junior engineers via 1:1s, code reviews and design reviews.

Led multiple company-wide DevOps initiatives like multi-region infrastructure, immutable infrastructure and infra-as-code tooling.

Tech stack: AWS, Ansible, Consul, Docker, Jenkins, Kong, Packer, Terraform.

• Maintaining and optimizing deployment pipelines.
• Leading Immutable Infrastructure efforts.
• Leading Infrastructure-as-Code efforts.
• Site reliability engineering and developer tooling.

Worked on Viki’s subtitling tools and platform.

Languages: JavaScript, Ruby.

Tech stack: Backbone.js, Docker, Drone, Graylog, Rails, RabbitMQ, Redis, PostgreSQL.

• Developed features across the full stack to improve the subtitling workflow.
• Refactored legacy code according to Clean Code principles.
• Eliminated N+1 database queries for highly-trafficked web pages.
• Conducted performance profiling and optimization of the subtitling tools to fix obscure memory leaks in older browsers.
• Improved telemetry by centralizing logging with Elasticsearch and introducing monitoring with Graylog.
• Performed root cause analysis using Graylog to diagnose and fix internal server errors, reducing them by 90%.
• Dockerized the local development environment, reducing the set-up time from days to minutes.

Developed a code formatter for the Elixir language. This project also involves open-source contributions to Elixir tokenizer and parser (written in Erlang) to make the abstract syntax tree more amenable for code formatting.

Mentors: Josè Valim and Andrea Leopardi.

Languages: Elixir, Erlang.

Test-driven & full stack development for Grab’s driver & passenger CMS.

Languages: JavaScript, Ruby.

Tech stack: React + Reflux, Rails, RSpec, Phabricator, TravisCI.

• Implemented features and fixed critical bugs in production across the full stack.
• Worked on multiple cities feature for drivers, allowing them to pick up jobs in different cities within the same country.
• Wrote a lot of unit tests using RSpec so as to adhere to test-driven development.

Automated manual labor components using computer vision to enable better operations.

Language(s): Python.

Tech stack: Airflow, Google Vision API, OpenCV, Tesseract.

• Researched and developed a prototype to extract structured data from receipts using OpenCV and Tesseract, achieving over 90% accuracy on in-sample dataset.
• Built and shipped a production-ready image processing pipeline by integrating the prototype with Google Vision API, speeding up order reconciliations by 3X and potentially saving thousands of dollars in operating costs per month.

Rebuilt the product across the full stack.

Languages: JavaScript (ECMAScript 6), Ruby, Swift.

Tech stack: React + Redux, Rails, Xcode.

• Rebuilt various parts of the product with HTML5, CSS3, ES6 using React and Redux.
• Built & shipped the entire chat SPA using socket.io to allow real-time updates, with responsive design across desktop and mobile platforms, shaving off response times by 80%.
• Rewrote legacy Rails API for new chat SPA to handle the shift of business logic to front-end, shrinking Rails codebase by 40%.
• Restructured CSS code using best practices such as BEM methodology.
• Developed features for iOS app using Swift and Xcode.
• Conducted code reviews and standardized coding style of the front-end team using style guides.

Improved Wego’s infrastructure by tinkering with its distributed systems.

Languages: Bash, Golang, Ruby.

Tech stack: AWS EC2, Chef, Consul, Docker, HAProxy, Heka, Jenkins, Kafka, Marathon, Mesos, Zookeeper.

• Wrote Chef recipes to automate infrastructure configuration.
• Developed a CI & CD pipeline using Jenkins to enable automatic deployments.
• Developed a container orchestration platform for staging environments using Marathon, Mesos and Docker.
• Implemented service discovery for the aforementioned platform using HAProxy and Ruby.
• Set up & maintained on-premise Docker registry which sped up Docker pushes & pulls by 5X.
• Developed a proof-of-concept for distributed configuration management using Consul.
• Set up and utilized Kafka as a message broker to make data pipelines fault-resilient.
• Developed & open-sourced a Heka plugin written in Go to persist data from Kafka to Amazon S3.

Made internal tools and worked on a project for Samsung APAC.

Languages: JavaScript, Ruby.

Tech stack: Angular.js, Rails, Amazon DynamoDB.

• Mainly worked on internal tools, such as employee portal, project management tool, and employee time tracker using Angular.js with Rails.
• Developed a scalable NoSQL message service API using Amazon DynamoDB for a client’s project (Samsung APAC).
• Migrated company’s internal database to Google Compute Engine.
• Maintained the backend infrastructure of the company’s in-house apps.