- 100M+category-page hits a day
- +18.07%GMB from one filter change
- −40%CPU, GraphQL vs REST
- $2.2Ma year, annualized from an ads test
Selected work
SEO landing pages at scale
- Problem
- Every one of eBay’s hundreds of millions of search-engine landing pages depends on URL rules that backend services generate and publish to production.
- What I built
- Designed, built and launched the microservices behind those pages, redesigned the URL library’s caching, and restored the publishing pipeline after a network change blocked every SEO, search and item-page team.
- Result
- Category pages alone average over 100M hits a day.
Sticky filter chips
- Problem
- On mobile web, the filter chips scrolled out of view as soon as shoppers reached the results.
- What I built
- Took the spike and the A/B test for chips that stay pinned while scrolling, then owned the launch to all mobile-web traffic and the code cleanup.
- Result
- GMB up 18.07% and transactions up about 15% in the top four markets.
Product pages on GraphQL
- Problem
- The team needed evidence that GraphQL could match the existing REST backend before moving product pages over.
- What I built
- Developed a GraphQL product-page backend on Spring Boot and Netflix DGS in about six weeks, then load tested it against REST.
- Result
- Same p95 latency as REST with 40% less CPU.
Cost-per-click ads test
- Problem
- Cost-per-click ads were blocked on category pages, and the ads team wanted to test them there.
- What I built
- Led the backend, writing the design, building click tracking, and finding and fixing a missing experiment header before the test ran.
- Result
- Ads revenue up about 5%, five times the forecast, roughly $2.2M a year annualized. Launch is on hold for a retest.
An agentic AI pipeline for shipping code
- Problem
- Every feature request turned into a spike document, then tickets, then code, then reviews.
- What I built
- Built a pipeline on Claude Code and a Jira MCP server that turns spike documents into tickets and takes each ticket to a pull request, with CLAUDE.md context files for each backend service.
- Result
- Every pull request of my final year at eBay shipped through it.
Experience
2025 — 2026
Member of Technical Staff 1 · eBay
San Jose, CA
Migrated the category-page listings grid to a GraphQL page platform across 10+ sites, built embedding-similarity retrieval for review highlights at a 99.75% SLO, and designed price ranges for trading-card variants that lifted conversion about 16%.
2023 — 2025
Software Engineer 3 · eBay
San Jose, CA
Led the backend for an ads test that raised ads revenue about 5%, moved a monolithic category-page backend onto a shared listings microservice, and delivered EU Digital Services Act compliance on time.
2020 — 2023
Software Engineer 2 · eBay
San Jose, CA
Launched sticky filter chips (+18.07% GMB), more than doubled filter usage on eBay Motors, and owned about 20 biweekly backend releases end to end.
2019
Software Engineer · Trade and Development Bank
Ulaanbaatar, Mongolia
Built backend and frontend features for internal banking systems and a credit-scoring system.
2017 — 2019
Full Stack Developer · Asterisk Technologies
Ulaanbaatar, Mongolia
Built and supported web applications and ERP projects for clients.
Education
- Master's, Computer Science, Maharishi University of Management, Iowa2019 — 2021
- Bachelor's, Software Engineering, National University of Mongolia2014 — 2018
About

I'm Bat. I started my career in Ulaanbaatar, Mongolia, building web apps and ERP systems at Asterisk Technologies while finishing a software engineering degree at the National University of Mongolia. I then built banking and credit-scoring software at Trade and Development Bank.
In 2019 I moved to Fairfield, Iowa, for a master's in computer science, and in 2020 I joined eBay. Over six years there I grew from Software Engineer 2 to Member of Technical Staff, working on the backend for category, search-landing and product pages.
The work I like most is where scale meets the shopper, like a latency budget, an A/B test, or a migration that has to land without anyone noticing. Lately that includes building AI agents into how code gets shipped.
Contact
I'm looking for my next senior software engineer, full-stack or AI platform role.