
Senior Software Engineer
Microsoft · Full-time
Feb 2024 - Present
• 1 yr 5 mos· Microsoft Worldwide Learning team, focusing on the AI Powered Authoring project(.Net C#, Python)
· Led the AI-Powered Authoring project, developing educational materials—including certification exams, practice/pre/summative assessments, PowerPoint presentations, and avatar-based videos—leveraging Large Language Models (LLMs) and Bloom’s Taxonomy to generate high-quality content.
· Optimizes AI-driven content generation using Retrieval-Augmented Generation (RAG), advanced prompt engineering, and Chain of Thought reasoning to enhance search accuracy and reduce hallucinations. Curates relevant documents with Azure Cognitive Search and OpenAI Embeddings, leveraging lexical and semantic similarity calculations to improve content relevance and quality.
· Ensures high-quality learning materials by defining evaluation guidelines and authoring rules. Utilizes AI-powered automated assessments for filter and iterative refinement, enhancing key metrics such as accuracy, relevance, clarity, effectiveness, and engagement to maximize learning impact.

Senior Software Engineer
Nextdoor · Full-time
Mar 2023 - Jan 2024
• 11 mos* Designed and developed the data structure and architecture of Nextdoor's Ad Server, optimizing its efficiency for better engineering. (Java, Python)
* Initiated an Ad Report feature to identify and rectify billing issues, ensuring client trust and retention by preventing accidental overcharges. (Kotlin)
* Implemented Price Floors and monitoring/logging systems to safeguard revenue and facilitate debugging. (Java, Python)
Skills: Apache · Amazon Web Services (AWS) · Python · Kotlin · Java

Senior Software Engineer
Meta · Full-time
Aug 2019 - Feb 2023
• 3 yrs 7 mos* Experienced with FAM Auction infra and E2E AdLoad (Backend/Web/Mobile) in the Facebook App Monetization (FAM) Ads Delivery team; solved real-world problems in auction and ad load.
* Led & initiated the “Unified Price Floor Library in Auction” project: developed “personalized reserve price” to increase the efficiency/accuracy of ads delivery; reduced free delivery by 98%; developed “cost per outbound click price floor” to enhance ads quality, reduced ad farm by 50%+, bringing $80M in annual profits.
* Designed and implemented the Dynamic Per Position Reserve Price Ad Load framework in the auction system for Feeds, bringing $40M each year.
* Unified auction framework across placements, improving synergy across all Meta products and reducing maintenance costs.
* Led the project E2E Dynamic AdLoad Control (Blueprint), increasing ads impression without sacrificing users’ engagement. Estimated annual profits: $300M.
* Co-led the Feed Ads Separation project: successfully separated Ads’ and Feeds’ data model; reduced the maintenance and communication cost between Feed & Ads groups
Skills: Objective-C · Python (Programming Language) · Apache Thrift · C++ · PHP

Software Engineer
Comcast · Full-time
Sep 2017 - Aug 2019
• 2 yrsWorked on the Advanced Advertisement Team, developed Audience and Campaign Explorer, which was a web-based explorer to help Linear and Operation Team conduct ad planning and evaluation; processed audience and campaign data using Spark.
Skills: Amazon Web Services (AWS)

Software Engineer
Mies Jansen
Feb 2017 - Jul 2017
• 6 mosDevelop, maintain and upgrade a system named “Crystal”: An app for Comprehensive Real Time Sales and Trading Analysis. The app is based on Meteor, React and MongoDB.
- Built a more secure infrastructure including Single Sign-On, Access Control, Data Encryption.
- Single Sign-On: Control login system by connecting Microsoft Service ADFS, based on SAML 2.0.
- Access Control: Provide different services based on roles of users.
- Data Encryption: Encrypt Sensitive and Private Data in Database.
- Added new features to increase Workflow Productivity, making Crystal faster, more convenient and
efficient in BWIC and Bond level, such as Custom Copy, external API.
- Rewrote and cleaned the code based on Model-View-Control architectural pattern.

Software Engineer Internship
MailTime
Jun 2016 - Aug 2016
• 3 mosData Analysis:
- Parsed emails and extracted conversation threads using Regular Expression.
- Employed Xpath to collect information (item name and price, item ID, etc.) from order
confirmation emails and electronic receipts for future data analysis.
- Used Hadoop to construct a big data architecture independently, which could upload email
information databases to HDFS storage on a daily basis.
Development and Testing:
- Developed Exchange Active Sync of Email Commands; tested them using Pytest.
- Tested Mailtime App Android Version to find any existing problems.