Pengxiang Zhang

About Me.

I am a AI Software Engineer specializing in the intersection of Generative AI Integration and Scalable Full Stack Architecture.

With 4 years of professional experience, I bridge the gap between traditional enterprise systems and modern LLM orchestration. Recently, I have spearheaded the development of enterprise-grade internal ChatGPT solutions, leveraging my full-stack background to build secure APIs, intuitive frontends, and robust backend orchestrations that serve thousands of users.

My primary focus is on Agentic Workflows. I engineered an intelligent GSS (Global Shared Services) Agent that automates complex corporate tasks by orchestrating connectivity between Workday, ServiceNow, and SAP. From architecting RAG-based knowledge systems with LangChain to implementing high-concurrency CI/CD pipelines on AWS, I am dedicated to delivering production-ready AI that solves real-world business challenges.

6 Years
Years of Experience
4 Years
Full Stack Exp
3 Years
AI Agents Built

Technical Stack.

Technologies I use to bring AI-powered full stack applications to life.

AI & Agentic Orchestration

OpenAI
Gemini
Anthropic
LangChain
MCP
AWS Bedrock
Vector Databases (Pinecone/Chroma)
TensorFlow

Primary Languages

Python
TypeScript
JavaScript
Go
Rust
C++
Java
C#
PHP

Full Stack Architecture

React
Next.js
Node.js
Tailwind CSS
PostgreSQL
MongoDB
Redis

Cloud, DevOps & Automation

Docker
AWS
GCP
Azure
Flask
Kubernetes
Git
CI/CD Pipelines

Education.

Northwestern University
09/2021 - 12/2022

Master of Science in Computer Science

Northwestern University

Specialized in Artificial Intelligence, Cloud Computing, and Scalable Systems.

University of Nebraska - Lincoln
08/2018 - 05/2021

Bachelor of Science in Computer Science

University of Nebraska - Lincoln

Focus on Algorithms, Data Structures, and Software Engineering.

Experience.

Syngenta

AI Software Engineer

SyngentaFull-Time
05/2023 - Present
  • Served as the Subject Matter Expert (SME) and Lead Developer for Generative AI, defining the company’s internal LLM roadmap and security standards.

  • Architected and spearheaded the deployment of a secure, enterprise-grade ChatGPT instance, ensuring zero-leakage of proprietary data while serving thousands of internal users.

  • Engineered a sophisticated RAG (Retrieval-Augmented Generation) framework using LangChain and Vector Databases, drastically improving organizational knowledge discovery.

  • Integrated an advanced AI Agent for GSS (Global Shared Services), orchestrating seamless connectivity between Workday, ServiceNow, SAP, and other core enterprise systems to automate complex employee workflows.

  • Architected and implemented robust CI/CD pipelines utilizing various AWS services (CodePipeline, CodeBuild) and GitHub Actions, ensuring rapid and reliable deployment of AI microservices.

  • Collaborated with cross-functional leadership to identify high-impact AI opportunities, reducing manual overhead by over 30% through custom automation tools.

Generative AIChatGPTLangChainVector DatabasesAWSCI/CD
Syngenta

IT Intelligent Automation Engineer

SyngentaInternship
06/2022 - 12/2022
  • Developed a large-scale Cloud Automation Solution using Gatsby (React + GraphQL), following best practices for scalable frontend architecture.

  • Leveraged AWS Glue to automate data processing tasks, utilizing serverless architecture to optimize scalability and minimize operational overhead.

  • Integrated multiple third-party APIs to create a unified automation workflow, enhancing cross-departmental efficiency.

  • Optimized system performance by identifying and resolving critical bugs, leading to improved application stability and user experience.

  • Conducted advanced troubleshooting for the automation pipeline, working closely with the DevOps team to maintain high system uptime.

GatsbyReactGraphQLAWS GlueGitGo
Northwestern University

Research Assistant & Developer

Northwestern UniversityInternship
06/2021 - 09/2021
  • Led a technical team of 6 in architecting an experimental workflow using React and Flask for a specialized legal chatbot; integrated Qualtrics and MTurk to automate and streamline large-scale data collection.

  • Implemented rigorous project management methodologies and coding best practices, ensuring the high-fidelity delivery of research software within tight academic deadlines.

  • Engineered a Conversational AI agent using Google Dialogflow and Landbot to provide automated legal guidance on commercial property issues.

  • Utilized Natural Language Processing (NLP) techniques to optimize intent recognition and entity extraction, significantly improving chatbot accuracy and user experience.

  • Designed and executed scientific research projects, applying data-driven techniques to explore complex legal questions and ensuring compliance with ethical AI guidelines.

ReactFlaskDialogflowNLPPythonProject Management
Aipaca Canada Inc.

Software Engineer Intern

Aipaca Canada Inc.Internship
06/2020 - 09/2020
  • Collaborated with a 4-person engineering team to develop a one-line model training CLI, abstracting complex infrastructure for data science teams using Agile methodologies.

  • Architected comprehensive software validation and system testing procedures, ensuring high-fidelity delivery and strict adherence to client requirements.

  • Engineered an adaptive privacy layer for legacy software and optimized the AWS infrastructure, resulting in a 300% acceleration in environment provisioning and setup time.

  • Implemented performance-driven modifications to ensure software compliance with evolving data privacy standards while maintaining system responsiveness.

AWSPythonCLI DevelopmentInfrastructure as CodeAgile
University of Nebraska - Lincoln

Full Stack Developer

University of Nebraska - LincolnInternship
05/2020 - 05/2021
  • Led a team of 3 to architect a high-performance Bioinformatics web platform using Flask (Python) to visualize research on polyphenol metabolism and human health.

  • Integrated specialized sequence analysis tools (BLAST and HMMER) into a server-side architecture, optimizing backend logic for large-scale biological data processing.

  • Deployed and maintained the production environment (bcb.unl.edu/dbpup/), ensuring enterprise-grade security, cross-browser compatibility, and accessibility.

  • Co-authored a published research article (DOI: 10.1128/AEM.01851-21) showcasing the platform's ability to drive future computational biology research.

  • Supervised the full migration of laboratory infrastructure to Amazon AWS, implementing cloud-native best practices for automated backups, disaster recovery, and institutional security compliance.

FlaskPythonAWSBioinformaticsDockerPostgreSQL
University of Nebraska System

IT Support Specialist

University of Nebraska SystemPart-Time
09/2019 - 05/2021
  • Provided technical support for a user base of 20,000+ students and faculty, resolving hardware and network issues with minimal downtime.

  • Developed a centralized Knowledge Base and technical documentation, reducing redundant support requests and improving team resolution speed.

IT OperationsTechnical DocumentationTroubleshooting
University of Nebraska - Lincoln

New Student Orientation Leader

University of Nebraska - LincolnPart-Time
09/2019 - 05/2021
  • Facilitated the onboarding of international students, managing complex logistics to support their transition into university life.

  • Devised data presentation schemes for sequence analysis outputs, helping researchers visualize complex datasets.

  • Represented the university at professional organizational meetings, demonstrating strong leadership and cross-cultural communication skills.

LeadershipPublic SpeakingData Visualization

Featured Work.

Airplane Boarding Model

Airplane Boarding Model

  • An agent-based model that uses Netlogo to simulate different boarding strategies on different size airplanes could reduce the time required for airplane boarding.
NetlogoAgent-based ModelComputer Modeling
Welcome To Game

Welcome To Game

  • Welcome To is a roll and write game with no dice or rolling involved.Instead, it uses a deck of cards. In this game, players build houses, group these houses into high-scoring estates, and improve the quality of their neighbourhoods with fancy pools and beautiful landscaping.
PythonUnit TestingData Structures
Hive

Hive

  • A community-focused app bringing together current students and alumni to connect, provide one another support, and stay engaged in the activity of the school community.
ReactFirebaseJavaScriptAgile
Photo App

Photo App

  • A community-focused app bringing together current students and alumni to connect, provide one another support, and stay engaged in the activity of the school community.
FlaskPythonReactJavaScriptCSSSEOUsability
UHangout

UHangout

  • A mobile app where allow students can find others to join them in group events.
ReactFirebaseJavaScriptAgile
Unity Game

Unity Game

  • 3D games build using Unity to experience game development, level design, world building and environmental storytelling.
UnityC#Level DesignWorld BuildingEnvironmental Storytelling
Learning Assistant Management System

Learning Assistant Management System

  • A one-stop website that can be used for Learning Assistant hiring, office hours, evaluations, and an on-demand appointment request system for students.
Ruby on RailsBootstrapAPIMySQL
Database for Polyphenol Utilized Proteins from gut microbiota

Database for Polyphenol Utilized Proteins from gut microbiota

  • dbPUP is a database website for a research project of polyphenol utilized proteins. It has 60 proteins with 24120 sequence displays on this website. Other than that it has a web interface server to run BLAST and HMMER.
FlaskBootstrapD3MySQL
Package Tracking Systems

Package Tracking Systems

  • A package tracking system automatically assign courier distribution stations and automatically plan routes to the nearest warehouse.
JavaGoogle Cloud PlatformHibernate
Socket Cypher Chat System

Socket Cypher Chat System

  • A simple 2-way chat program, essentially a bi-directional echo server/client. Only one socket is set up, so only two chatters are possible. Use DES Cipher, Base64 Cipher and Pain text to encrypt the chat.
JavaCypherSocket
Book Management System

Book Management System

  • Implement part of a simple library book management system. Most of the functional requirements come from user stories prepared by students.
JavaMavenUser Story

Let's Connect.

Whether you have a question about AI Agents, LLM Architecture, or just want to say hi, my inbox is always open.

📍 Chicago, IL, USA

📍 Shenzhen, Guangdong, China