AI expertise, on demand.
A senior AI lead embedded in your team 10–20 hours a week. Fixed-fee sprints that put a working pilot in production — not a slide deck.
ex-Meta data science · €560M national data platform · 1M+ users shipped
Nexus Sprint Board
AI Build Sprint · Support Copilot
- W1
Audit & eval set
- W2
RAG pipeline v1
- W3
Pilot behind flag
- W4
Handover
eval · answer accuracy
86%
target ≥ 85% · 8 eval runs
hours this week
14 / 20 h
deliverables
- Architecture doc
- Golden eval set
- Runbook
// how it works
Embedded enough to own outcomes. Flexible enough to stay senior.
You don't need a 40-person consultancy or a contractor waiting for tickets. You need someone who has shipped AI at scale, sitting in your standups.
Use-case selection, build-vs-buy, model and vendor choices, and the evals that decide what ships.
Now
Support copilot
Next
Churn-risk scoring
Later
Agentic onboarding
10–20 h / week
A fixed slice of a senior lead. Capped at 2–4 clients, so you're never the fifth priority.
Ships production code
Pilots go live behind a feature flag in your stack — OpenAI, Vercel, Node.js, your cloud.
if (flags.supportCopilot) { draft = await copilot(ticket) }
Fixed fee, not hourly
Sprints are priced by outcome. You know the cost before day one.
$12,000/ 4-week build
Mentors your team
// the ai build sprint
Four weeks. One working pilot. One fixed fee.
$12,000
fixed · no hourly surprises
- 01
Week 1
Architecture & evals
Data and infra audit, target metric, golden eval set, architecture doc signed off.
- 02
Week 2
Core pipeline
Retrieval, prompting or fine-tune path built against your real data. First eval run.
- 03
Week 3
Pilot in production
Shipped behind a feature flag to real users, with logging, guardrails and cost tracking.
- 04
Week 4
Handover
Eval results, runbook, roadmap for v2 and pairing sessions with the engineers who own it next.
// why not the usual options
Built for the gap the others leave open.
Marketplaces sell hours. Matching services sell introductions. Consultancies sell decks. Nexus sells a pilot in production and a team that can run it.
Nexus
Senior operator
- Owns outcomes & AI strategy
- Pricing
- Fixed-fee sprints
- Time to first pilot
- 4 weeks
- Ships production code
- Mentors your team
- Embedded 10–20 h / week
Toptal
Talent marketplace
- Owns outcomes & AI strategy
- Pricing
- Hourly
- Time to first pilot
- Depends on you
- Ships production code
- Mentors your team
- Embedded 10–20 h / week
- Hourly blocks
Go Fractional
Matching service
- Owns outcomes & AI strategy
- You manage scope
- Pricing
- Hourly / placement
- Time to first pilot
- Depends on match
- Ships production code
- Varies
- Mentors your team
- Embedded 10–20 h / week
Accenture & boutiques
Consultancy
- Owns outcomes & AI strategy
- Via SOW
- Pricing
- Six-figure minimums
- Time to first pilot
- Months
- Ships production code
- Often slides
- Mentors your team
- Embedded 10–20 h / week
// scope your sprint
Describe the feature. Get a draft sprint plan in seconds.
Tell us what you're piloting or scaling. You'll get a 4-week plan, the risks that matter, and the sprint we'd start with.
plan.preview
Your 4-week plan appears here.
A week-by-week timeline, concrete deliverables, the three risks most likely to sink it, and which sprint we'd recommend.
Week 1
Week 2
Week 3
Week 4
// who you work with
Saverio Pulizzi
Founder, Nexus · Warsaw
Ex-Meta data scientist, former tech lead of a €560M national data platform, and a founder who has shipped GenAI products to over a million users. He has lived and worked across seven countries.
- €560M
- EU funding secured via the national AI strategy
- 140M
- monthly API requests at AdsOn.ai
- 1M+
- users on Neural Canvas products
- 100+
- senior officials trained on GenAI
2025 –
CEO, Denovo
AI multi-agent platform for starting a company. Raised $750,000; 1,500 live projects.
2024 – 25
Co-founder & CEO, AdsOn.ai
Scaled to 140M monthly API requests. Nvidia Inception, Mondadori PLAI accelerator.
2020 – 24
Data Science Expert, Italian Prime Minister’s Office
Tech lead of a €560M Data Platform; National Data & AI Strategy; G7 and OECD AI roadmaps; led a 20+ person team.
2021 – 23
Balance Consulting
GenAI strategy for a major bank, leading to a $500K initiative.
2016 – 20
Data Scientist, Meta
NLP to scale product support globally; prevented millions of fake accounts per month; a KPI adopted company-wide.
2020 –
Founder, Neural Canvas Inc.
Early GenAI products used by 1M+ users: neuralcanvas.io, chartai.io, aigify.com.
Awards 2025: Best AI-Powered Comic Creation Platform · Digital Comics Disruptor of the Year.
// pricing
Priced by outcome. Published up front.
Start with a sprint. Stay on as your fractional AI lead if it works — most teams do.
Discovery Sprint
For teams still deciding what to build.
$6,000
2 weeks · fixed fee
- Use-case validation
- Data & infra audit
- Build-vs-buy recommendation
- Roadmap + eval plan
AI Build Sprint
For teams ready to ship a pilot.
$12,000
4 weeks · fixed fee
- Architecture & model choices
- Working pilot in production behind a flag
- Evals you can trust
- Team handover & runbook
Fractional AI Lead
For teams scaling an AI product line.
from $8,000
per month · 10–20 h / week
- Owns AI roadmap & outcomes
- Leads implementation
- Mentors your engineers
- Limited to 2–4 concurrent clients
Retainer work blends to roughly $200/hour — near the $223 fractional average — billed as one monthly fee.
// faq
Questions CTOs ask first.
Something else? Join the waitlist and ask on the first call.
How is this different from hiring through Toptal?
Toptal gives you a vetted contractor billing by the hour; you still own the strategy and the outcome. Nexus is a senior AI lead who owns both, on a fixed fee.
What do we need to have in place before a Build Sprint?
A use case you want to test, access to the relevant data, and one engineer who will own the system afterwards. If the use case isn’t validated yet, start with the 2-week Discovery Sprint.
Do you write code or just advise?
Both. The Build Sprint ends with a working pilot in production behind a feature flag, plus the architecture doc, evals and runbook.
Why only 2–4 clients at a time?
Because 10–20 hours a week only works if you get real attention. The cap keeps every engagement senior.
Which stack do you work in?
Your stack. Typical builds use OpenAI and other model providers, Node.js, Vercel and the major clouds, with evals from day one.
What happens after the sprint?
Your team keeps everything. Many teams continue with the Fractional AI Lead retainer to scale what the sprint proved.
// next sprint slots open monthly
Book an AI Sprint.
Leave your work e-mail. We'll reply with available start dates and a 30-minute scoping call.
2–4 clients at a time · no spam, one e-mail when a slot opens