Portrait of Nielzon Tolentino

Hi. I'm Nielzon Tolentino, an applied AI engineer.

And I am an Applied AI Engineer. I architect LLM agents, serverless backends and AI-powered products for healthcare and fintech, from multi-agent clinical form autofill to a policy-gated LLM customer support pipeline, and own them from design through production.

Learn about what I do

Here's all the stuff I do.

Unlocking Digital Potential, One Line of Code at a Time.

AI Agents & LLM Systems

Production AI:
Architect multi-agent and LangGraph systems on Claude, OpenAI and AWS Bedrock, with Databricks model serving, RAG over Databricks Vector Search and Weaviate, structured outputs, prompt caching, and replay and grounding evaluations that gate every model change.

Serverless & Event-Driven Backends

Built to Scale:
Engineer Go services on AWS Lambda with DynamoDB Streams, SQS and Databricks, plus Laravel queues and multi-tenant APIs, designed around idempotency, backpressure and clear ownership of every event.

Security & Reliability

Hardened by Design:
Close replay and data-exposure gaps in service-to-service auth, sanitize sensitive data out of logs, mask PII before it reaches a model, and back every fix with tests and alerting.

Full-Stack Product Engineering

End to End:
Ship React, Next.js and Node products with real-time dictation, Stripe metered and team billing, and SSO, from the first schema to the release pipeline.

Here's some stuff I've worked on recently.

Recent creations to inspire you.

Professional work

NavixScribe clinical documentation pipeline

NavixScribe

The AI clinical scribe behind navix.ai. I engineered the pipeline end to end: chunked audio upload with persistence and retry, Deepgram dictation behind an ephemeral-token broker, template-driven note generation, and a LangGraph benefits-verification agent with per-case memory. A delta output contract cut form-autofill output tokens by 98%.

SmartFill multi-agent autofill pipeline

SmartFill

Multi-agent form autofill for the Navix Hub behavioral-health EMR, which I architected and led. An orchestrator plans field groups, runs agents in parallel over targeted chart retrieval, and merges schema-validated, PII-filtered output across OpenAI and AWS Bedrock, replacing one long serial prompt per form.

Monee AI support pipeline architecture

Monee AI Support

A serverless AI customer-support backend for an e-commerce payments platform, which I built as its primary engineer: Go on AWS Lambda with DynamoDB Streams and SQS, Claude on Databricks with retrieval, a policy gate on every auto-reply, and a replay harness that gates model changes.

Have me make stuff for you.

Tell me about your project, and let's work together to turn your digital dreams into a reality. Your vision, my expertise, a perfect match.

Email me at: nielzon.t@gmail.com


Find me on ...