Luca Palonca · Software engineer, Italy
I build and fix the systems your product runs on.
Backends, data pipelines, and the AI features that sit on top of them — built to hold up in production, not just in a demo. Delivered async as pull requests against your repo.
No pitch, no pressure. Describe what you're building or what's breaking, and within two business days you'll get a written read on it — including if the honest answer is that you don't need me.
I'm Luca — 7+ years a software engineer, currently Backend Tech Lead on a US healthcare SaaS platform. Across those years I've migrated 51M records with zero downtime, taken an API from 8 seconds to 350ms, and more recently built guardrails and evals over 2M+ clinical documents with 3.5M+ LLM requests under cost and rate-limit control. More about me →
How I can help
- Build it
- One well-defined thing, designed and shipped — a service and its API, a data pipeline, a migration that can't take the platform down, or an AI feature that has to be reliable enough to put in front of users.
- Fix it
- A backend that's become slow, expensive or fragile, or an AI feature you can't trust — audited to a written report that ranks what's actually wrong, plus pull requests fixing the top of the list.
- Decide it
- A data model, a vendor, a migration approach, a service boundary — reviewed before you commit to something painful to undo, by someone who has made the call before and will tell you what they'd actually do.
Selected work
A real-time voice AI agent, in production, in healthcare
Building a real-time conversational AI agent with multi-modal audio streaming, tool calling, and structured outputs for a regulated healthcare platform.
A 51M-record migration with zero downtime
Designing a batched, multi-threaded data migration pipeline that moved 51M+ records without taking the platform offline.
Stopping an LLM fabricating clinical facts across 2M+ documents
How grounding constraints, schema-validated extraction and continuous evals kept an LLM summarization pipeline from inventing details across millions of clinical documents.
Free tools
No signup, nothing to install, nothing leaves your browser.
Why is my RAG hallucinating?
A free diagnostic for production RAG systems giving wrong, unsourced or fabricated answers. Answer what you're seeing, get a ranked list of likely root causes with the check that confirms each one.
LLM Cost Calculator
Compare per-request and monthly API costs across Claude, GPT, and Gemini models for your actual token usage.
Golden set size calculator
How many eval examples do you actually need to detect a regression? Enter your golden set size and find out the smallest accuracy drop it can reliably catch.