Mark Bao Consulting
Cellar
Wine inventory management system
Design, frontend/backend development, process strategy
Cellar is an inventory management system, custom-built for a commercial wineseller. Cellar manages inventory across multiple locations using a barcode label system, and includes an iPhone app that scans barcodes to view and check out items.
I worked with the client to design the barcode label according to their specifications, and meet their inventory control workflow by including features like Inventory Check.
Designed and developed from scratch, working with the client to design a process for labeling inventory items and a check-in/check-out flow.
Atom
Financial data and news platform
Design and development of full product, data ingestion and management, real-time chat
Atom is a source of financial data and news for investors. It provides real-time equity price data, a place to track stocks you’re interested in, and detailed historical financial data. To help investors find promising equities, Atom provides a screener for finding new stocks based on 20 financial indicators. Finally, it allows groups of investors like investing clubs to collaborate with a group chat system.
I designed the product to feel forward-thinking and innovative, differentiating it from current financial data tools, using a periodic table-like ‘stock tile’ format to represent companies, which gives the user a quick way to view the performance of a stock.
- Designed and developed a responsive, single-page application (SPA) frontend with real-time stock price updates, using React and GraphQL
- Developed a backend and API with user accounts, and stock tracking, with Node.js, TypeScript, and PostgreSQL
- Implemented a system to ingest data for tens of thousands of ticker symbols and hundreds of thousands of financial reports into a normalized format, and allowing users to search for companies by specifying criteria for 20 financial indicators
- Designed and developed a real-time group chat system with stock ticker auto-completion based on Apollo GraphQL
- Instrumented the app for analytics using Segment and Mixpanel
ArXivLab
Academic research discovery platform
Backend development, machine learning, server architecture, some frontend
ArXivLab is an interface to arXiv, an academic e-print service with over one million papers. ArXivLab includes research discovery features like personalized recommendations and machine learning-based article similarity models. It is developed at the Blei Lab at Columbia University, where I was a research assistant.
- Developed and deployed two machine learning models (word embeddings, topic modeling) for article similarity over the entire 1.3 million article database, with the data pipeline built in Python, frontend in React, and API in Node.js
- Contributed to Custom Views, a feature for researchers to segment the arXiv database to tailor their literature search, using React and Node.js
- Managed and optimized the migration of 1.2 billion data points into MySQL
- Developed and optimized an article popularity system using MySQL and Node.js
- Redesigned the server architecture using Docker
Kindred
Referral recruiting software
Design and development of full product and marketing pages, data ingestion and management, payment integration
Kindred is a recruiting tech product that helps companies hire from their existing employee networks. It allows employees to opt in to allow recruiters to search for candidates that existing employees are already connected to, automating a process that’s happening manually within companies today, or not at all.
- Designed and developed a single-page application (SPA) frontend, along with responsive marketing and pricing pages, using React and GraphQL
- Developed a backend and API with user accounts, permissions, employee inviting, and user management using Node.js, TypeScript, and GraphQL
- Optimized PostgreSQL database queries for finding matching candidates for a position
- Implemented a data processing queue to process user-given data
- Implemented payment integration to allow customers to subscribe to the application using Stripe