I work with cloud infrastructure, production systems, monitoring, incident response, deployments, and operational reliability.
I'm currently expanding that foundation into Applied AI and Forward Deployed Engineering, with a focus on building systems that interact with real business data, use tools, retrieve evidence, expose their execution, and solve practical problems.
My current areas of focus include:
- AI agents and tool calling
- Retrieval-Augmented Generation (RAG)
- Semantic and hybrid retrieval
- PostgreSQL + pgvector
- FastAPI-based AI applications
- Agent observability and execution traces
- Evidence grounding and hallucination control
- Production-minded testing and reliability
AI Customer Support Resolution Agent
Support Pilot AI is a tool-using customer-support agent built around a fictional SaaS platform called CloudDesk.
Instead of relying only on an LLM's internal knowledge, the agent can choose approved tools, inspect structured business data, retrieve support documentation, and generate responses grounded in evidence.
V1 — Agent Foundation ✅
- Gemini tool calling
- Explicit agent orchestration loop
- Customer and subscription lookup
- Service incident investigation
- Semantic knowledge retrieval
- Gemini Embedding 2
- PostgreSQL + pgvector
- Evidence-grounded responses
- Hallucination guardrails
- Developer View / Agent Trace
- Agent-run and tool-execution persistence
- Tool latency and failure tracking
- 28/28 automated tests passing
Next: V2 — Multi-Tool Resolution
Multi-user, multimodal RAG knowledge workspace
Knowledge Hub AI evolved across three versions from a document-based RAG application into a retrieval-focused knowledge platform with private user workspaces, multimodal document understanding, hybrid retrieval, grounded generation, and broader evaluation.
Key capabilities
- PDF, DOCX, Markdown and TXT ingestion
- PostgreSQL + pgvector
- Gemini embeddings and generation
- Hybrid semantic + lexical retrieval
- Candidate ranking / reranking
- Grounded answer generation
- Source attribution
- Conversational retrieval
- Authentication and private workspaces
- Admin / member roles
- Cross-user data isolation
- Multimodal document understanding
- Flowchart, chart and architecture-diagram retrieval
- Unsupported-question rejection
- 107 automated tests passing
AWS · CloudWatch · Linux · Terraform · Docker
CI/CD · Jenkins · Git · Bitbucket · Octopus Deploy
New Relic · CloudWatch · PagerDuty · Incident Response · Production Support
Python · FastAPI · PostgreSQL · SQLAlchemy · SQL · pgvector
Gemini · LLM Tool Calling · AI Agents · RAG · Embeddings · Semantic Search · Hybrid Retrieval · Grounding
pytest · API Testing · Workflow Testing · Evaluation · Regression Testing
What interests me about Forward Deployed Engineering is the combination of technical problem solving, customer context, system integration, production ownership, and measurable outcomes.
The kind of workflow I enjoy looks like this:
Business / Customer Problem
↓
Understand the Requirement
↓
Investigate Systems & Data
↓
Build or Integrate a Solution
↓
Validate the Behavior
↓
Observe & Troubleshoot
↓
Improve the Outcome
My Cloud/DevOps background gave me experience around production systems, reliability, monitoring and incident response.
My current AI projects are helping me build the other side of that skill set: agents, retrieval, tool use, data integration, grounding, evaluation and AI application design.
Knowledge Hub AI
├── V1 ✅
├── V2 ✅
└── V3 ✅
Support Pilot AI
├── V1 — Agent Foundation ✅
└── V2 — Multi-Tool Resolution → Next
I'm intentionally building these projects version by version so that each release adds a deeper engineering problem rather than simply increasing the feature count.
- AWS Certified Solutions Architect – Professional
- AWS Certified Cloud Practitioner
I'm interested in opportunities around:
Forward Deployed Engineering · Applied AI · AI Agents · Cloud / DevOps · Technical Solutions Engineering
Building at the intersection of Cloud Engineering, Applied AI, and real-world problem solving.