Predictable GCP bills · Production-grade AI · Self-sufficient teams
You don't need to chase every tech trend to scale. I help C-level leaders make the bets that matter, design GCP architectures shaped to your reality, ship purposeful AI from PoC to production, and build engineering teams that don't need a babysitter. Peace of mind, by design.
> i run a handful of engagements at a time. grab 20 minutes, tell me what's on fire in the booking notes, and we'll see if it's in my wheelhouse.
>> nicola@guglielmi ~ / SYSTEM_MODULES
>> MODULE_01.SH
[STATUS: OK]
Strategic Tech Leadership
Hiring elite talent, mentoring Tech Leads, and building engineering teams that ship, motivated, accountable, no babysitting required. Org topologies, DevOps culture, and the operational discipline to outlast me.
>> MODULE_02.SH
[STATUS: OK]
C-Level Strategic Advisory
A second brain for CTOs, founders, and exec teams. Which tech trends to ride, which to ignore, when to invest, when to wait. Decisions backed by architecture and pattern-recognition, not by hype cycles.
>> MODULE_03.SH
[STATUS: OK]
Scalable GCP Architectures
Designed to your shape, scale-to-zero where it fits, distributed where it doesn't, boring monolith where it just works. Terraform, GKE, Cloud Run, BigQuery. Costs predictable, services decoupled, ownership transferred to your team.
>> MODULE_04.SH
[STATUS: OK]
R&D + AI Prototyping
From idea to PoC to production-grade MVP, time-boxed sprints with explicit kill criteria, no science projects. Vertex AI, RAG with grounded citations, agentic workflows that actually pass the hallucination bar.
>> nicola@guglielmi ~ / ENGAGEMENTS
Consulting work is usually under NDA . I can't name clients. I can show you the patterns I keep solving, the shape of the engagement, and the outcomes you can expect.
NDA · STRATEGIC · FINANCIAL · CRITICAL-INFRA
>> PATTERN_01[ 6-10 WEEKS ]
FinOps Rescue
GCP bills go from unpredictable to flat
Node-pool right-sizing, spot for stateless workloads, killing zombie environments, commit-discount strategy, and the playbook your team uses after I leave. No vendor lock-in to a SaaS FinOps tool.
→ TYPICAL OUTCOME:
30-45%
bill reduction · 6-10 weeks
sectors: SaaS · fintech · media
>> PATTERN_02[ 4-12 WEEKS ]
GCP Architecture & Migration
from legacy hosting to GCP-native, zero downtime
Target architecture design, cutover plan with explicit kill criteria, traffic-shift in waves, rollback ready at every step. Scale-to-zero where it fits, distributed where it doesn't, monolith where it just works.
→ TYPICAL OUTCOME:
−60%
infra spend · zero downtime
sectors: media · fintech · national-strategic
>> PATTERN_03[ 6-12 WEEKS ]
AI Integration (PoC → Prod)
purposeful AI in production, hallucination bar enforced
Use-case selection (kill the wrong ones early), grounded RAG with citations, eval harness for regression, cost + latency budget enforced, team handover with playbook. No ChatGPT-wrapper energy.
→ TYPICAL OUTCOME:
In prod
grounded, eval-gated, team-owned
sectors: legal-tech · healthcare-adjacent · public
>> PATTERN_04[ 4-6 WEEKS ]
Strategic Advisory (C-Level)
second brain for the bets that matter
Sparring partner for CTO / VP Eng / founder on technology bets: build vs buy AI, GCP vs hybrid, team topology pre-Series-B, M&A tech due diligence. Prep doc → live conversation → written recommendation → 30-day follow-up.
→ TYPICAL OUTCOME:
3-5
sessions · written recs · follow-up
sectors: scale-ups · post-IPO · PE portfolio
>> PATTERN_05[ 8-16 WEEKS ]
Team Topology & Mentoring
org shape + tech leads that don't need a babysitter
Current-state interviews, topology redesign (platform + stream-aligned), role definitions, tech-lead coaching plan, hiring funnel calibration, DevOps culture transfer. The goal is a team that runs without me, and without anyone like me.
→ TYPICAL OUTCOME:
25+
eng team restructured · 1 quarter
sectors: scale-ups in growth phase
> patterns repeat; specific engagements don't. references on request, never published.
>> nicola@guglielmi ~ / LABS
Twelve months heads-down building. Eight MVPs shipped, across conversation, multimodal, audio, text, dev-ops, B2B SaaS, and geospatial, most of them AI-native, to learn the craft end-to-end. These aren't client work. They're the receipts on the sabbatical year I spent making sure I know what I'm talking about when you hire me to ship the real thing.
[ CONVERSATION · WEB ]
[ LIVE ]
TheNegotiator
negotiation practice against lifelike AI counterparts
Next.js + Genkit + Gemini. Each AI counterpart runs its own persona and difficulty through Genkit flows, with structured-output scoring grading every round. Fully playable offline via scripted fallback, live with an API key. Gamified curriculum (XP, streaks, leagues), multilingual (EN/IT), Firebase auth with per-uid isolation, Cloud Run deploy.
Cloud Scheduler fires a Cloud Run Job daily: Gemini summarises the day's AI news, Vertex AI TTS voices it, cover art is rendered, the episode ships to YouTube and Spotify. Runs in Italian and Spanish, containerised, idempotent, zero-touch operation.
Next.js 16 + Genkit. One flow chains Gemini 3.5 Flash for the narrative into Gemini 2.5 Flash Image for matching illustrations. Firebase auth + Firestore for the personal story shelf, deployed on Firebase App Hosting. Quiet, brutalist UI; reCAPTCHA v3 against abuse.
GenkitGemini 3.5Image GenFirebase
[ TEXT · SAAS ]
[ LIVE ]
GrantBrain
multi-tenant SaaS, match grants, co-draft proposals
FastAPI + React. Document AI extracts, Vertex AI powers the RAG over funding calls, and Gemini-based semantic search matches grants to the applicant. Row-level multi-tenancy in Postgres, resumable Playwright crawler for funding portals, three subscription tiers and self-healing schema migrations.
Next.js dashboard fusing DORA delivery metrics with SPACE-framed human signals, plus a live tech-debt index (dependency freshness, rework ratio, change-failure) computed straight from git and GitHub webhooks. Role-shaped views for Maker / Manager / CTO with k-anonymity to protect individuals. Dual-driver SQLite to Cloud SQL Postgres, deployed on Cloud Run.
Django 5 + DRF + HTMX. Real-time safety scoring across deadlines, inspections, and equipment compliance (D.Lgs 81/08). NFC + QR attendance in the field, full audit trail with granular rollback, Stripe-billed multi-tenant SaaS on Cloud Run.
Sentinel-2 to operational agronomy, pixel to answer
FastAPI + Leaflet. On-demand Sentinel-2 (Copernicus) analysis: NDVI/NDMI with SCL cloud masking, satellite-derived field boundaries (region-growing), live cadastre via WMS/WFS, and a 3D vigour twin. Dual-mode from mock to real ESA data.
> 8 MVPs from one sabbatical year. the consulting work that pays the bills lives under NDA, see ENGAGEMENTS above.
>> nicola@guglielmi ~ / SCOPE
I take on a handful of engagements at a time. That only works if we're a fit, so here's the filter, up front.
GOOD FIT[STATUS: OK]
Scaling on GCP and your bills have gone unpredictable.
You want purposeful AI in production, not another demo.
10+ engineers, real budget, a real timeline.
You want to own your stack, not rent a pair of hands.
NOT A FIT[STATUS: SKIP]
Pre-revenue side projects with no budget.
"Build my MVP" staff augmentation.
Non-cloud, mobile-only, or crypto work.
Shopping for the cheapest contractor.
>> nicola@guglielmi ~ / HOW_I_WORK
Three phases. Each is time-boxed with explicit deliverables. No open-ended retainers, no permanent dependency on me, the goal is to leave the system in your team's hands.
>> STEP_01[ 1-2 WEEKS ]
Diagnose
find what's actually broken
I dig into your stack, your bills, your team, and the trends you should care about. No PowerPoint. No "discovery workshops." Just reading code, asking hard questions, and looking at the numbers.
→ OUTPUT: written diagnosis with the 3 things worth fixing first
>> STEP_02[ TIME-BOXED ]
Architect
design the target and the path to it
I design the target system, the migration path, and the team shape needed to get there. Decisions explained, trade-offs explicit, alternatives killed on purpose. Executable, not aspirational.
→ OUTPUT: a plan your team can run, with milestones + kill criteria
>> STEP_03[ UNTIL DONE ]
Hand-off
transfer ownership, kill the dependency
I work alongside your team as architect and coach, pair-programming the hard bits, reviewing PRs, mentoring leads. Explicit exit criteria. No retainer creep.
→ OUTPUT: your team owns the system. zero key-person risk left behind.
> sometimes the engagement stops at step 01 (diagnosis is enough). that's a feature.
GOOGLE DEVELOPER EXPERT
GOOGLE CLOUD AUTHORIZED TRAINER
RESIDENT TECH PARTNER @ SAMNIUM HUB
> this site uses google analytics to measure traffic. cookies are set only if you accept. details in the privacy policy.