Backend & AI/Agentic engineer building SaaS platforms and the agents behind them
Node.js and PostgreSQL/MongoDB on the backend side, LLM orchestration and RAG pipelines on the AI side — schema design, agent architecture, and the problems that only show up once real usage hits.
Where most of the work actually goes
A few things come up on nearly every project: keeping the API layer sane, getting the database schema right early, not finding out about a bottleneck in production, and — increasingly — giving an LLM agent enough guardrails to be trusted with real actions.
APIs that don’t block
Express services that hand off slow work to a queue instead of making the request wait for it.
Schemas that hold up
Indexing and query patterns thought through up front, in both PostgreSQL and MongoDB — tenancy boundaries included.
Endpoints worth trusting
Validated inputs, versioned routes, and auth that’s actually been tested against the roles it’s supposed to restrict.
Agents that don’t run wild
Tool-calling loops with schema-validated inputs, guardrails on what they can touch, and a memory layer that won’t quietly blow the context window.
The stack, in practice
What's actually in use across current projects — not an exhaustive list, just what gets reached for.
Node.js
ExpertEvent-driven, asynchronous runtime for scalable network systems.
Express.js
ExpertFast, unopinionated, minimalist web framework for RESTful APIs.
TypeScript
ExpertStatically typed superset of JavaScript ensuring compile-time safety.
REST API Design
ExpertBuilding semantic, versioned, rate-limited, and scalable API structures.
A few things I've built
Case studies with the actual problem, the architecture, and what shipped — not just screenshots.
ROI Spectrum
The Problem: SEO agencies and website owners lacked an integrated, multi-tenant solution to track ROI, audit logs, service tracking, and automated subscription billing transparently.
Forklift CRM
The Problem: Operations managers struggled to handle lead assignments, track workflow automations, and manage sales representatives with clean role-based permissions.
AgentCore
The Problem: Every LLM provider ships its own SDK, message format, and streaming API, so switching models, or falling back to a second provider when one is rate-limited, usually means rewriting the integration layer.
Memorybook
The Problem: Tools like NotebookLM are useful for asking questions over your own documents, but they are closed products, no self-hosted option, no visibility into how retrieval or memory decisions actually get made.
GitHub activity
What the commit history actually looks like, day to day.
Open Source Activity
Simulating daily commit pushes, branches, and code reviews
What working together looks like
Sample feedback, illustrative of the kind of work above — not direct quotes from a specific reviewer.
"Handed off the ROI Spectrum schema and Stripe webhook work and didn't have to think about it again — billing edge cases included."
"The CRM's lead-assignment logic needed real-time updates and a permissions model that wasn't a mess. Both landed clean on the first pass."
Have an API or database problem?
Happy to talk through it — whether that's a new build, a slow query, or a SaaS backend that needs to hold up under real usage.