I build production web applications and enterprise software — and I'm currently
going deep on AI systems (RAG, agentic workflows) on top of that foundation.
I'm a full-stack developer who enjoys taking a product from a vague requirement to a deployed system that real users depend on. Most of my work lives on the backend and architecture side — auth, authorization models, data design, service boundaries — paired with React/Next.js front-ends that make those systems usable.
Recently I've been focused on applied AI: integrating LLMs into existing products, building retrieval-augmented generation (RAG) features, and learning how to design agentic workflows responsibly inside production software rather than as demos.
- 🏗️ I think in terms of systems and trade-offs, not just features.
- 🔐 I care about correct authorization (RBAC/ABAC), audit trails, and data integrity.
- 🤖 I'm actively leveling up in AI engineering — RAG, embeddings, and tool-using agents.
- 🐧 I'm comfortable in Linux, Docker, and the full deployment lifecycle.
Growing from "developer who integrates AI" into "engineer who designs AI-native systems."
- RAG pipelines — chunking, embeddings, retrieval quality, and grounding LLM output in real data.
- Agentic workflows — tool-using agents that automate multi-step business processes.
- Gemini API integrations — structured generation and content automation inside live products.
- DevOps fluency — containerized deployments, CI/CD habits, and observability.
Enterprise platform for capturing and managing clinical study data — regulated, audit-heavy, and isolation-first.
A modular CDMS built for an environment where data integrity and traceability are non-negotiable. The interesting engineering here is less about CRUD and more about governance: who can do what, proving it later, and keeping every customer's clinical data fully isolated. Instead of a shared multi-tenant database, each customer is shipped their own dedicated instance — the strongest isolation guarantee, and the expectation for regulated clinical systems. Audit trails and electronic signatures are designed toward 21 CFR Part 11 expectations.
- Dynamic Form Builder — configurable data capture (eCRFs) without redeploying.
- RBAC / ABAC authorization — fine-grained, attribute-aware access control.
- Audit trails & electronic signatures — every change attributable and verifiable (21 CFR Part 11).
- Single-tenant, isolated deployments — a dedicated instance per customer for full data isolation.
- Workflow automation over studies and user management with a modular service design.
Next.js · Node.js · PostgreSQL · RBAC/ABAC · Single-tenant deployments
Commercial drone-service marketplace connecting partners with customers.
A two-sided platform where I built out partner onboarding, the service catalog, and the first AI-assisted features — using the Gemini API to generate service descriptions and reduce manual content work.
- Partner management & service marketplace with scalable backend services.
- AI-generated service descriptions (Gemini API) feeding the catalog.
Languages JavaScript · TypeScript · SQL
Frontend React.js · Next.js · HTML/CSS
Backend Node.js · Express.js · REST APIs · Microservices
Databases PostgreSQL · MySQL · MongoDB
Auth & Security OAuth · JWT · RBAC / ABAC · Single-tenant data isolation
AI & Automation (Learning) RAG systems · Gemini API · LLM integration · Agentic workflows
Infrastructure Docker · Linux · CI/CD · System Design
How I approach building software:
| Area | What it means in my work |
|---|---|
| System Design | Clear service boundaries, sensible data models, and trade-offs I can defend. |
| Authorization | Treating access control (RBAC/ABAC) as a first-class part of the design. |
| Deployment Isolation | Shipping dedicated single-tenant instances so each customer's data is fully isolated. |
| Auditability | Building systems where actions are traceable — audit trails, signatures, history. |
| Scalability | Designing backends that grow without a rewrite. |
I treat AI as an engineering discipline, not a feature checkbox. My current focus:
- RAG — retrieval quality, chunking strategy, and grounding answers in source data.
- LLM integration — wiring the Gemini API into real product flows (e.g. AerialBorne).
- Agentic workflows — letting tool-using agents handle multi-step processes reliably.
- Honest scope — I'm learning fast and shipping; I describe what I've built, not hype.
I keep a deliberate learning track alongside shipping work:
- 📚 Now: RAG architecture, embeddings, and agent design patterns.
- 🛠️ Next: Deeper DevOps — orchestration, monitoring, and infrastructure-as-code.
Open to roles and collaborations in full-stack engineering, SaaS, and applied AI.