Zuo Wang

Staff AI Engineer

Email: amangoworks@gmail.com | Phone: +1-757-633-7384
Location: San Diego, US (places I've lived in)

Staff-level AI backend infra engineer with 6 years of professional experience and $10M+ business impact. Cross-disciplinary technical breadth in Software & Hardware Engineering, ML, Security, Product, and Infrastructure.

People Leadership: Experienced people leader, grew team to 7 people. Fun to work with, appreciate smart engineers with low ego Interview Experience: Conducted dozens of algorithm and system design interviews Track Record: Proven record in strategizing and delivering reliable systems at scale through cross-team collaborations


Technical Expertise

Languages & Frameworks

Golang Python gRPC GraphQL

Cloud & Infrastructure

GCP AWS Kubernetes VertexAI

AI/ML

vLLM Claude Pinecone Arize

Data & Observability

MongoDB Grafana

Work Experience

Samaya AI, London

Member of Technical Staff

Jan 2024 - Current
  • Architect Knowledge Ingestion Pipeline, Embedding Chunking strategy, Eval, RAG, and AI Infrastructure
  • 0-1 turn ML ideas into product features like QueryDecomposition, DeepResearch, and EarningsReport
  • 90% performance improvement: Decrease time-to-first-token from 100s to 10s, scale infra from 10 to 10,000 daily active users
  • Onboard customers from Morgan Stanley, BlackRock, Citi, and Point72. Strategize pricing
  • Demo for our $43.5M Series A

Company Website

Tesla, Shanghai GigaFactory + Texas GigaFactory

Staff Software Engineer & Team Lead, High Voltage Manufacturing

Jul 2022 – Jan 2024
  • Responsible for high voltage systems: Model 3/Y/S/X, Powerwall, Megapack, and SuperCharger
  • Iterate chip design for CyberTruck and Megapack under tight deadline (2.5 months) and heavy pressure
  • Close collaboration with worldwide vendors (CATL, BYD, TI, COSMX, HYC, Pegatron, Jabil). Travel to their countries for design reviews, on-site inspections, troubleshooting, and reducing defect rates
  • Teach teams in Manufacture, Process, and Quality to enable them to do their own debugging and fault analysis. Made China battery production line 3x faster than USA line
  • Finish CyberTruck battery charger 4 weeks ahead of schedule
  • Bring up 50 new manufacture stations, maintain over 200 stations, handle $500k budget per year

Atlassian, Mountain View, CA

Software Engineer, Cloud Security + Platform Infrastructure

2018 – Jul 2022 | Fully Remote
  • Develop auth sidecar, secure for all communication between Atlassian microservices and staff
  • Rewrite tools like LDAP cache, rate limiter, and deduper from Java, TypeScript, and Python to Golang
  • On call and answer questions for a 99.99% availability internal PaaS hosting over 1,400 Atlassian services
  • Create instant generator for services with bells and whistles like K8s, SOX compliant, and CI/CD

Tools: Splunk, SignalFX, Jira, Confluence, AWS S3, DynamoDB, Lambda, Postgres, Spinnaker


Education

Johns Hopkins University

Master of Science, Machine Learning Security

GPA: 3.9/4.0 | 2018

Machine Learning Security: Cloud Security, Software Vulnerability Analysis, Computer Forensics, Network Security, Privacy Laws, Risk Management, Ethical Hacking, Modern Cryptography

University of Rochester

Bachelor of Science, Electrical & Computer Engineering

GPA: 3.6/4.0 | 2014

Joseph C Wilson “Change” Scholarship • Dean’s List

ECE: Signal Processing, Robot Control, Semiconductor, Microprocessor, C++, Java, Assembly, VHDL, MATLAB Minor: Mathematics and Economy


Notable Projects & Research

Research Experience

Behavioral Biometrics Security Research - Unknot.id

Security Research Intern | Aug-Dec 2019
- Researched and implemented adversarial attacks against Android biometric authentication systems - Developed efficient shadow model attacks requiring fewer than 1,000 queries to compromise security - Applied defensive hardening techniques to LSTM models using the CleverHans adversarial learning framework - Published research findings on behavioral biometric vulnerabilities [Research Publication](https://www.unknot.id/post/yes-you-can-spoof-behavioral-biometrics-with-adversarial-learning)

Cognitive Security Modeling - Johns Hopkins University

Graduate Research Assistant | Jan-Aug 2019
- Developed computational models in Python to simulate human security decision-making processes - Implemented advanced cognitive architectures including Instance-Based Learning Theory and ACT-R memory systems - Processed and analyzed 40GB+ of behavioral datasets to understand security-related cognitive patterns - Contributed to interdisciplinary research bridging computer science and behavioral psychology [Research Profile](http://behavior.isi.jhu.edu/people.html)

Technical Projects

Ontra.ai

- Replace thin GPT wrapper Ruby on Rails software to proper ML service that does retrieval, cross-encoding, prompt evaluation - Reduce thousands of hours for human-in-the-loop lawyers wasted on poor AI output [ontra.ai](https://www.ontra.ai)

Agemo Codewords

- Fullstack webapp for automated website building with NodeJS and Shadcn [agemo.ai/codewords](https://www.agemo.ai/codewords)

Rippling Global Expansion Framework

- Build out global expansion framework for Rippling's payroll product using Python Django, MongoDB, and React

LeetCode

- Solved 1000+ questions, improved contest rating from 1460 to 2109 in 7 months [leetcode.com/mangoman](https://leetcode.com/u/mangoman/)

Military Drone System - U.S. Army Contract

- Designed and manufactured drone-based supply drop system for military training operations - Utilized 3D CAD design, additive manufacturing, and Arduino-based control systems - Deployed at National Training Center and Joint Readiness Training Center for operational use

Battle Map Gaming Platform

- Built real-time multiplayer gaming platform with WebSocket-based interactions - Implemented full-stack architecture using modern web technologies for scalable gameplay - Deployed cloud-hosted solution with real-time synchronization capabilities [Live Demo](https://battle-map-344101.wl.r.appspot.com/)

Security Research & Cybersecurity

  • Docker Privilege Escalation: Applied Dirty COW race condition for container-to-host escape
  • IoT Exploitation: Network analysis and attacks on DJI drones, Bebop drones, Amazon Echo
  • Malicious URL Detection: ML models with PhishTank and DMOZ datasets, ROC curve analysis
  • Software Vulnerability Analysis: Threat modeling and reverse engineering with objdump, ldd, nm, strace
  • CryptoDoneRight: Contributed to cryptodoneright.org server hardening tools

Additional Interests

Gaming & Software Development

  • Gaming platform development
  • Top 0.3% League of Legends player

Music & Performance

  • Guitar player with live concert experience
  • University carillon society member

Athletics & Fitness

  • Standup paddleboard enthusiast
  • Amateur boxing
  • 10+ pull-ups personal record

Electronics & Hardware

  • Headphone amplifier design
  • Windshield wiper controller
  • Custom PC builds