AI Systems Engineer | Evaluation-First Agentic AI | RAG, Data Pipelines, Automation, and Infrastructure
I build practical AI and software systems that turn messy operational data into reliable, measurable, human-accountable workflows.
My current focus is evaluation-first agentic AI: systems that retrieve grounded context, reason over it, evaluate their own outputs, and either improve, escalate, or stop when reliability is low.
I come from a full-stack, infrastructure, and cybersecurity background, with experience across Java, Python, React, AWS, automation, CI/CD, and enterprise engineering. Today, I am applying that foundation toward AI systems that can actually survive real-world use: retrieval pipelines, agent orchestration, evaluation frameworks, workflow automation, and domain-specific intelligence tools.
The next wave of useful AI will not be won by demos alone.
It will be won by systems that can be:
- Grounded in real data
- Evaluated with clear metrics
- Observable under failure
- Corrected through feedback loops
- Deployed into real workflows
- Held accountable to human judgment
That is the direction behind my current work in RAG, agentic refinement, workflow automation, and applied AI systems.
A from-scratch conversational RAG and evaluation framework focused on retrieval quality, grounded generation, failure analysis, and measurable reliability.
Current interests include:
- Conversational retrieval
- Query rewriting
- Context precision and recall
- Faithfulness and answer relevance
- Cost, latency, and reliability tradeoffs
- RAG as a trust layer for downstream AI workflows
A private R&D effort exploring agentic refinement, self-improving workflows, control levers, and AI systems that can evaluate and improve their own behavior over time.
The long-term goal is not simply to build agents, but to build systems that know when to accept, retry, refine, escalate, or stop.
A data engineering and AI pipeline project focused on structured weather data, Bronze/Silver/Gold data layers, reproducibility, and downstream intelligence applications.
A vertical AI concept for weather-aware operational intelligence, risk analysis, and decision support.
Applied automation and AI-assisted systems for real businesses, including lead workflows, websites, operational tooling, and measurable client outcomes.
Academic and technical work involving structured data, language, research support, and applied AI/data workflows.
- Python
- RAG pipelines
- Embeddings and vector search
- LLM evaluation
- Conversational AI
- Agentic workflows
- Data pipelines
- SQL
- Structured dataset design
- Java
- JavaScript
- TypeScript
- Node.js
- React
- Spring
- Spring Boot
- REST APIs
- Full-stack application development
- AWS
- Docker
- GitHub Actions
- Ansible
- Linux
- CI/CD
- OpenShift
- Firebase
- Networking
- NAS / remote infrastructure workflows
- SAST / DAST
- Vulnerability remediation
- Secure Java applications
- CSRF protection
- Infrastructure hardening
- Observability-minded system design
- Client systems
- Workflow automation
- Lead-generation tooling
- Website rebuilds
- Technical consulting
- Requirements translation
- Outcome-focused engineering
A responsive and accessible website for a nonprofit organization, built with React and Node.js. Includes event-oriented and donation-oriented functionality.
A conversational RAG evaluation project focused on building and explaining retrieval-augmented generation systems from scratch.
Core themes:
- Retrieval quality
- Grounded answer generation
- Conversational context handling
- Evaluation metrics
- Failure-mode analysis
- Research-oriented documentation
A weather data pipeline and AI infrastructure project focused on structured data processing, reproducibility, and downstream intelligence.
A private agentic AI research and engineering effort focused on refinement loops, system control, reliability, and model/workflow improvement.
A client-facing workflow and business system focused on automation, lead handling, and measurable real-world value.
I am working toward becoming a systems-level AI engineer who can bridge:
- Research and production
- Data pipelines and decision systems
- LLMs and real business workflows
- Automation and human accountability
- Infrastructure and applied intelligence
- Technical depth and ethical responsibility
My goal is to build AI systems that are not just impressive, but useful, reliable, measurable, and aligned with human flourishing.
I care about building systems that are:
- Truth-seeking — grounded in data, evidence, and clear evaluation
- Human-accountable — especially in high-stakes workflows
- Non-harmful — technology should not be used to dehumanize or destroy
- Maintainable — useful systems must survive beyond the first demo
- Measurable — if a system cannot be evaluated, it cannot be trusted
- Practical — real users, real constraints, real outcomes
- Faith-shaped — my work is guided by stewardship, truth, humility, and responsibility
- Conversational RAG evaluation
- Agentic AI system design
- LLM reliability and failure modes
- AI workflow orchestration
- Data engineering for AI systems
- Cloud and local compute workflows
- AI infrastructure for rapid experimentation
- Research writing and reproducible technical artifacts
I'm interested in serious projects involving:
- RAG systems
- AI evaluation
- Agentic workflows
- Vertical AI applications
- Workflow automation
- Data pipelines
- Applied AI for small businesses
- Faith-aligned or values-driven technology
- Human-accountable AI systems
- RAG and conversational AI
- AI system evaluation
- Full-stack engineering
- Java / Spring Boot
- React / Node.js
- AWS and infrastructure automation
- Workflow automation
- Secure application development
- Turning messy operational processes into software systems
- B.S. in Mathematics, Computer Science, and Physics — Augsburg University
- Graduate work in Artificial Intelligence
- AWS Certified Cloud Practitioner
- Former infrastructure and software engineering experience across enterprise environments, including financial services, aviation, retail, and software development.
- Portfolio: dbraun.io
- LinkedIn: linkedin.com/in/david-braun777
- Email: davidjbraun777@gmail.com
- GitHub: github.com/DavidBraun777
I am fluent in sign language and care deeply about using technology to serve people, strengthen communities, and build systems that create real value.