AI Engineer | Full Stack Tech Lead Dallas, TX

Building intelligent software. Exploring the future of AI.

Maimaiti Aimaier. I'm a full stack tech lead with nine years of experience shipping secure APIs and cloud-native services in banking, healthcare, and logistics. Now I'm applying that production discipline to agentic AI: multi-agent systems, tool calling, and MCP.

Explore My Projects
Foundation
  • Java 21 · Spring Boot
  • React · Next.js
  • AWS · Kubernetes
  • Secure APIs · CI/CD
Now building
  • LLM agents
  • Tool calling
  • MCP servers
  • Evaluation
Portrait of Maimaiti AimaierAI twin available

Ask my AI twin

An AI representation of me, grounded in my documented background.

  • 100+

    CVEs closed in platform-wide remediation

    PNC Bank

  • 2 mo → ~2 wk

    to deliver new AML detection scenarios

    PNC Bank

  • 40%

    fewer deployment-related incidents

    BNSF Railway

  • 0

    findings across multiple compliance audits

    Abbott Laboratories

About

An enterprise engineer, now building agents.

I build the layer other teams depend on: reusable frameworks, secure APIs, and cloud-native services. Over nine years that has meant regulated environments in banking, healthcare, and logistics, where security, auditability, and release discipline are part of the definition of done.

At PNC Bank I own the architecture, API contracts, security model, deployment strategy, and production support for a Java 21 / Spring Boot extension layer on NICE Actimize, supporting anti-money-laundering detection. That is also where AI entered my day job: I built an LLM-powered agent and MCP server for compliance analysts, with access controls and audit logging, and I led AI enablement for engineers: usage guidelines and security review of AI tools.

Alongside that work I have been going deep on agentic AI. I completed the AI Engineer Agentic Track (Udemy), earned the Agentic AI certificate from DeepLearning.AI, and built multi-agent systems with the OpenAI Agents SDK and MCP. This site is the most complete example: an AI twin of me, grounded on a structured profile, with automated checks for what it should and should not say.

Bachelor of Computer Science, Xinjiang University, Urumqi, China. Relevant coursework: Machine Learning, Algorithms, Database Systems, Software Engineering, Web Database Applications.

How I work

  • Build what other teams build on

    I favour reusable foundations over one-off features.

    Evidence: A reusable authentication and authorization framework adopted by multiple integration teams at PNC; a React component library adopted by multiple product teams at Abbott.

  • Security is part of delivery

    Controls belong in the pipeline and the architecture, not in a late review.

    Evidence: Automated security scanning in CI/CD and platform-wide CVE remediation (100+ CVEs closed) at PNC; access control and data protection that passed compliance audits with zero findings at Abbott.

  • Make AI agents production-ready

    An agent that touches regulated data needs access controls, audit logs, and clear guidance, not just a good prompt.

    Evidence: An LLM-powered agent and MCP server for compliance analysts with access controls and audit logging; AI usage guidelines and security review of AI tools at PNC.

Experience

Nine years of shipping in regulated industries.

Banking, healthcare, logistics, and telecom: platforms where security, auditability, and release discipline are part of the job.

  1. –

    Current role

    Full Stack Tech Lead Engineer · PNC Bank

    Financial crime / anti-money-laundering (AML) detection platform built on NICE Actimize

    • Own architecture, API contracts, security model, deployment strategy, and production support for a Java 21 / Spring Boot extension layer on NICE Actimize, supporting compliance analysts and transactions.
    • Built an LLM-powered agent and MCP server for compliance analysts to detect fraud cases, with access controls and audit logging for a regulated environment.
    • Led AI enablement for engineers: usage guidelines, security review of AI tools, and partnership with security, architecture, and compliance on responsible-AI adoption.
    • Applied AI coding assistants (Claude Code, GitHub Copilot) in the team's engineering workflow for coding, test generation, refactoring, CVE fixes, and code review.
    • Built a reusable authentication and authorization framework for REST APIs and batch processes, adopted by multiple integration teams.
    • Designed Spring Boot microservices that extend Actimize detection models.
    • Integrated automated security scanning into CI/CD pipelines and led platform-wide CVE remediation, including transitive vulnerabilities in vendor components.
    • Partner with compliance teams, analysts, product owners, and downstream application teams to turn detection requirements into secure, production-ready services.

    Impact

    • Cut delivery of new AML detection scenarios from vendor release cycles of about 2 months to about two weeks.
    • Closed 100+ CVEs and reduced late-stage remediation effort by 30%.
    • Improved release quality and compliance readiness through automated security validation and proactive vulnerability management.
    • Java 21
    • Spring Boot
    • NICE Actimize
    • REST APIs
    • MCP
    • LLM agents
    • CI/CD
    • Claude Code
    • GitHub Copilot

  2. –

    Technology Lead · BNSF Railway

    Modernization of critical rail operations platforms

    • Led modernization of a critical rail operations platform, defining the target architecture, migration strategy, engineering standards, and implementation roadmap.
    • Established code review and architecture governance across distributed teams and mentored offshore Java and React engineers.
    • Re-architected legacy applications into containerized microservices on Kubernetes and OpenShift, improving scalability and throughput metrics.
    • Designed CI/CD pipelines with GitHub and Jenkins that automated build, test, and deployment.

    Impact

    • Increased release frequency from quarterly to monthly and reduced deployment-related incidents by 40%.
    • Transformed a critical legacy system into a cloud-native platform, streamlining operations and supporting long-term scalability.
    • Java
    • React
    • Microservices
    • Kubernetes
    • OpenShift
    • Jenkins
    • GitHub

  3. –

    Full Stack Developer · Abbott Laboratories

    Enterprise healthcare applications on AWS microservices

    • Built a reusable React component library adopted by multiple product teams, standardizing the user experience.
    • Owned end-to-end API development in a distributed AWS microservices architecture, delivering REST APIs for React front ends and integrated healthcare systems handling thousands of requests per day.
    • Implemented access control and data-protection mechanisms for regulated healthcare data.
    • Standardized service interfaces to reduce integration effort for consuming applications, and translated business requirements into technical specifications.

    Impact

    • Reduced UI development time by 30% through the shared component library.
    • Passed multiple compliance audits with zero findings.
    • Reduced integration effort for consuming applications through reusable services and standardized APIs.
    • React
    • AWS
    • Microservices
    • REST APIs

  4. –

    Full Stack Developer · Verizon

    • Built Spring Boot microservices on AWS integrated with React front ends; supported cloud deployments and Agile sprint planning with product owners.
    • Spring Boot
    • AWS
    • React
    • Microservices

  5. –

    Full Stack Developer · Citigroup

    • Built secure web applications with Spring Boot and Angular, designing and consuming REST APIs for cross-system data integration.
    • Spring Boot
    • Angular
    • REST APIs

Technical skills

Depth where it counts, and honest about the frontier.

Core strengths come from production work. AI engineering is hands-on and growing; areas still being explored are labelled as such.

AI Engineering

Hands-on · growing

  • LLM APIs (OpenAI, Anthropic Claude, Google Gemini)
  • Multi-agent orchestration (OpenAI Agents SDK, CrewAI, LangGraph, AutoGen)
  • Model Context Protocol (MCP)
  • Tool / function calling
  • Prompt engineering
  • Structured outputs (Pydantic)
  • Gradio, Hugging Face Spaces
  • Amazon Bedrock (this site)

Backend Engineering

  • Java 21
  • Spring Boot
  • Python
  • Node.js
  • Kotlin
  • REST APIs
  • GraphQL
  • Microservices
  • Kafka / event-driven architecture
  • Batch processing
  • API authentication and authorization

Frontend Engineering

  • React
  • TypeScript
  • JavaScript
  • Next.js
  • Angular
  • HTML5 / CSS3
  • Reusable component libraries

Cloud and DevOps

  • AWS (EC2, ECS, EKS, Lambda, S3, RDS)
  • Azure
  • Docker
  • Kubernetes
  • OpenShift
  • Jenkins
  • GitHub
  • Maven / Gradle
  • CI/CD with automated security scanning

Data

  • PostgreSQL
  • MySQL
  • Oracle
  • MongoDB
  • SQL

Security and Practices

  • OWASP
  • CVE remediation
  • Regulated environments (AML / financial crime, healthcare)
  • Architecture governance
  • Agile / SAFe / Scrum

Actively exploring

Exploring
  • Retrieval-augmented generation (RAG)
  • AI application evaluation

Listed as areas of exploration rather than established production experience.

Projects

Practical AI systems, built end to end.

Agents, MCP, and a full-stack AI application in production. Details come from what I built; nothing here is padded.

Featured · Live

AI Digital Twin (this website)

An interactive AI representation of me, grounded on a structured profile and deployed on AWS with infrastructure as code.

Problem: Let visitors ask questions about my background and work, and get answers grounded in documented facts instead of reading a static page.

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • FastAPI
  • Python
  • AWS Lambda
  • API Gateway
Live (opens in a new tab)GitHub (opens in a new tab)

Request path

  1. Browser
  2. CloudFront + S3
  3. API Gateway
  4. Lambda (FastAPI)
  5. Amazon Bedrock

Infrastructure managed with Terraform and deployed with GitHub Actions. Model credentials stay on the server side.

  • Conversational agent · Built

    Career Digital Twin Agent

    A conversational agent deployed on Hugging Face Spaces that answers recruiter questions about my background.

    • Python
    • OpenAI API
    • Tool calling
    • Gradio

  • Multi-agent pipeline · Built

    Multi-Agent Deep Research System

    A planner, search, and writer agent pipeline that produces a structured research report and emails it.

    • Python
    • OpenAI Agents SDK
    • asyncio
    • Pydantic

  • Multi-agent system with MCP · Built

    Agentic Trading Simulation with MCP

    A four-agent trading simulation where agents use multiple MCP servers to research, decide, and execute simulated trades.

    • Python
    • OpenAI Agents SDK
    • MCP
    • Gradio

AI engineering journey

What I'm learning, building, and taking away.

A factual log rather than a progress bar: courses I've finished, work in progress, and lessons that come from the code in this repository.

Completed

  • AI Engineer Agentic Track: The Complete Agent & MCP Course

    Udemy (Ed Donner). Completed.

  • Agentic AI

    DeepLearning.AI certificate, issued Aug 2026 (listed on LinkedIn).

In progress

  • Generative and Agentic AI in Production

    Udemy course, in progress.

  • Retrieval-augmented generation and AI evaluation

    Active areas of exploration. This site uses a structured knowledge file and an offline evaluation suite rather than RAG, because the profile is small enough to fit in the prompt.

Building

  • This AI digital twin

    Grounded knowledge, safer request handling, observability, and an evaluation harness. See the project write-up.

  • AI agents for regulated environments

    At PNC: an LLM-powered agent and MCP server for compliance analysts, with access controls and audit logging.

Lessons from this build

  • Keep knowledge in data, not in prompt strings

    The twin's knowledge lives in one structured JSON file shared with the website, so a fact is corrected in one place.

  • Test the refusals, not just the answers

    The evaluation cases include missing information, invented achievements, and prompt-extraction attempts, and the default suite runs offline at no model cost.

  • Log behaviour, not conversations

    Operational logs record latency, token usage, and versions, but never message contents or secrets.

Current focusMaking AI agents secure, reusable, and production-ready.

AI Digital Twin

Don't just read about me. Ask.

This is an AI representation of me. It answers from my documented background, explains technical topics, and tells you when it doesn't know.

Omar's AI twin

AI representation, grounded in my documented profile

AI twin says:

Hi, I'm the AI twin of Omar (Maimaiti Aimaier). Ask me about my engineering background, my projects, or how I'm moving into AI engineering. I'm an AI built from my documented profile, so I'll tell you when something isn't covered.

Try asking

AI-generated from my documented profile. It can be wrong, so verify anything important.

Grounded in a structured profile

Answers come from one verified JSON profile, the same file that renders this site. The whole profile fits in the prompt, so there is no vector database to maintain.

Honest about gaps

If something isn't documented, the twin says so instead of improvising. It separates my documented experience from general technical explanation.

Tested like software

Offline tests and a separate live evaluation suite check grounding, missing information, invented achievements, and prompt-extraction attempts.

Messages are sent to Amazon Bedrock and stored server-side so follow-up questions have context. Please don't share sensitive information.

Contact

Let's talk.

The twin can cover the basics. For roles, collaborations, or anything that needs a human, reach me directly.

Based in Dallas, TX.