// backend · data pipelines · ai-agent systems

Dino Meschini

I build backend and AI-agent systems in Go and Python on Azure.

One example is a provenance-enforced equity-analysis pipeline over SEC EDGAR/XBRL filings, with typed review gates, fail-closed required inputs, and live end-to-end validation on AAPL, MRNA, and UBER.

Open to backend roles — remote, Argentina-based, EU work authorization (Italian citizenship).

Projects

3 shipped, solo
01
finance-skills-hermescase study →

Multi-agent LLM pipeline over SEC EDGAR/XBRL filings, with cross-model review of every generated analysis.

role   solo builder
stack  Python · Pydantic · EDGAR/XBRL · multi-agent LLM
status portfolio piece
02
Antesala

WhatsApp-native intake agent in Go; LLM-based triage ran live with real users before the product was wound down.

role   solo builder
stack  Go · WhatsApp API · LLM intake
status wound down · 2026
03
Oravela

Full product — Go/PostgreSQL backend, React/Vite front end, Claude-assisted document processing — built and operated on OCI end to end.

role   solo builder
stack  Go · PostgreSQL · React/Vite · OCI · Claude API
status wound down · 2026

Experience

2 selected roles
Hapi

Led the migration of legacy microservices from JavaScript to Go and built gRPC APIs backed by PostgreSQL and MongoDB.

role   Backend Developer
period Oct 2025 - Apr 2026
stack  Go · gRPC · PostgreSQL/MongoDB · Kubernetes · AWS
Humand

Automated onboarding operations with Python scripts and API integrations for reliable data exchange.

role   Customer Onboarding Operations
period Mar 2025 - Jul 2025
stack  Python · API integrations · operational automation

Engineering approach

provenance

Every number a system emits should be traceable to its source. If you can't audit it, you can't ship it.

end-to-end ownership

From schema to deploy to the invoice for the infra. Three products taught me what "done" actually costs.

direct communication

Plain statements of what works, what doesn't, and what it will take. No marketing energy.