Open to Senior / Staff opportunities
Hi, I’m Wei Jiang
AI Products
Senior Software Engineer with 14 years building scalable B2B SaaS platforms, distributed systems, and AI-native products — from Stripe payments to Uber mobility to Perplexity’s agentic AI research platform.
ScrollAbout
Engineering AI-native products at scale
Senior Software Engineer with 14 years of experience architecting scalable B2B SaaS platforms, distributed systems, and AI-native products. Delivered high-impact systems across Stripe payments, Uber mobility infrastructure, and Perplexity agentic AI platforms. Skilled in building intelligent applications that combine robust backend architectures, cloud platforms, and modern AI technologies.
0+
Years of Experience
$0B+
Annual Payment Volume Enabled
0M+
Daily Trips Powered at Uber
0%+
Citation-Grounded Accuracy
Experience
14 years across payments, mobility & AI
From Stripe's payments core to Uber's real-time marketplace to Perplexity's agentic research platform.
Agentic AI Research Platform · Multi-Agent AI Workflows · Enterprise RAG & Retrieval · 20+ Foundation Models
90%+
citation accuracy
20+
foundation models orchestrated
100M+
AI requests / month
Built an enterprise-grade Agentic AI Research Assistant platform that extended Perplexity's AI search capabilities with autonomous research workflows, secure data integrations, and production-scale LLM infrastructure.
- Architected and delivered a production-scale Agentic AI Research platform using Python, Java, AWS Bedrock, MCP and a custom agent orchestration runtime, enabling multi-agent workflows for planning, retrieval, analysis, and citation-grounded answer generation.
- Built an Orchestrator-Worker agent framework enabling dynamic task planning, intelligent model routing, recursive agent execution, context management, and autonomous recovery for complex, long-running research workflows.
- Developed an advanced AI search and RAG platform using Vespa, hybrid retrieval, semantic ranking, and citation validation — combining lexical search, vector retrieval, and multi-stage ranking to achieve 90%+ citation-grounded answer accuracy.
- Implemented an MCP-based tool execution layer connecting AI agents with enterprise applications, APIs, databases, and knowledge sources through standardized interfaces, secure authentication, and controlled context exchange.
- Engineered an AWS Bedrock-powered LLM platform with model routing, prompt management, guardrails, and inference optimization, orchestrating 20+ foundation models to balance response quality, latency, and cost.
- Built LLM evaluation and observability infrastructure using RAGAS, LangSmith, and production telemetry to analyze agent trajectories, retrieval quality, citation accuracy, faithfulness, latency, cost, and task success.
- Designed scalable backend services and event-driven AI pipelines using Python, Java, C++, FastAPI, Kafka, Vespa, PostgreSQL, Redis and AWS, enabling asynchronous agent execution and reliable data persistence.
- Developed key AI application interfaces and API services using React, TypeScript and Node.js for interacting with autonomous research workflows and agent execution states.
- Built and operated cloud-native infrastructure on AWS EKS with container orchestration, autoscaling, observability, and reliability engineering to support hundreds of millions of AI search and inference requests per month.
Real-Time Marketplace Systems · 40M+ Daily Trips · 1M+ Location Events/sec · ML-Powered Decision Systems
35%
dispatch failure reduction
15%
ETA accuracy improvement
1M+
location events / sec
Payments Infrastructure · Financial Systems · API Platform · Transaction Processing
$20B+
annual payment volume
99.9%
duplicate-failure reduction
4x
reconciliation speedup
Featured Work
Systems built to run at scale
A closer look at three platforms that shaped how their companies operate.
Perplexity AI
Agentic AI Research Platform
An orchestrator-worker multi-agent framework powering autonomous research workflows — dynamic planning, model routing, recursive execution, and citation-grounded answer generation across 20+ foundation models.
90%+
citation accuracy
20+
models orchestrated
Uber
Marketplace Intelligence Platform
Real-time dispatch and matching infrastructure processing over a million location events per second to power ETA prediction, demand forecasting, and sub-100ms rider-driver matching at global scale.
40M+
daily trips
-35%
dispatch failures
Stripe
Payments Infrastructure Platform
Distributed, idempotent payment processing services underpinning authorization, capture, refunds, and reconciliation for tens of billions of dollars in annual merchant volume.
$20B+
annual volume
100K+
merchants served
Skills
The toolkit behind the work
Languages
Backend
Frontend
AI Technology
Database
Cloud / DevOps
Architecture
Observability
Other
B.S., Engineering Mathematics and Statistics
University of California, Berkeley · 2008 — 2012
Contact
Let’s build something intelligent together
Based in San Diego, CA. Open to Senior and Staff engineering roles focused on AI products, distributed systems, and full-stack platforms.
Say hello