Talent, Enablement & AdvisoryAI Engineering Talent
FDE Sprint
In eight weeks, BigStep forward-deployed engineers work inside the client’s team to put one AI system into production, and leave the client’s own engineers able to change and redeploy it without us.
FDE Sprint
$40K–$100K
- Duration: 8 weeks
- Tier: Embed
- Format: Managed capacity
You leave withA deployed AI system and structured knowledge transfer to the client’s team
Book a Free SessionSound familiar?
We need this live this quarter, and we don’t want a black box when the agency leaves. Our engineers are good, but none of them has put an LLM system into production before.
What you leave with
System in production
Format: Client’s environment
How we know it’s acceptableMeets the production-ready definition agreed in week 1.
Evaluation set
Format: Client’s repository
How we know it’s acceptableRe-runnable by the client’s team without us.
Architecture document
Format: Document
How we know it’s acceptableReviewed and accepted by the client’s tech lead.
Runbook
Format: Document
How we know it’s acceptableA named client engineer has followed it to redeploy the system.
Knowledge-transfer checklist
Format: Signed document
How we know it’s acceptableEach named engineer has made and deployed a change unaided.
Also included
- Weekly pairing between our engineers and the client’s named engineers.
- A final handover session.
How it runs
8 weeks
Week 1
Scope, architecture, success criteria, knowledge-transfer plan.
OutA one-page spec and the agreed definition of production-ready.
Weeks 2–4
Build, paired.
OutA working version in staging by the end of week 4.
Weeks 5–6
Hardening, evaluation, monitoring; client engineers lead the changes.
OutEvaluation baseline and monitoring in place.
Week 7
Production deployment.
OutThe system live.
Week 8
Handover.
OutRunbook, signed knowledge-transfer checklist, handover session.
Is this right for you?
This is for you if
- Buyerthe engineering leader who has a specific AI system to ship and wants their team to learn by building it.
- Also has to say yesthe team lead whose engineers will pair with ours, because their time is committed for eight weeks; security; Finance above the threshold.
This is the wrong answer if
- The client has no engineers to receive the knowledge. Sell the Full Stack pod or the AI Feature Sprint instead.
- The client’s engineers cannot commit pairing time. The knowledge transfer is half of what they are buying; without it, the cheaper Feature Sprint is the honest answer.
- The system is not defined, or the data is not accessible. Start with the AI Opportunity & Roadmap Sprint.
- The work clearly needs more than eight weeks. Split it into two sprints, or use an FDE.
What’s included, and what isn’t
In scope
- one system
- one production environment
- knowledge transfer to the named engineers
- documentation, evaluation set and monitoring
Out of scope
- a second system
- data cleanup
- changes to unrelated systems
- support after handover (Managed Services)
- training anyone beyond the named engineers
Before we start
Before day one
- Named client engineers with committed pairing time.
- Repository, environment and data access.
- Representative sample data, not a cleaned subset.
- A production environment and a known path to get deployment approved.
- Agreed success criteria.
What we’ll need from your team
- Two named engineers, eight to ten hours a week each, pairing with ours.
- A domain expert, about two hours a week, to say whether the system’s answers are right.
- A platform or DevOps engineer, about two hours a week, for access and deployment.
- An approver available within 24 hours.
Pricing
What we need to give you a quote
- A description of the system.
- Data access.
- How many client engineers will take part, and their available hours.
- The target environment and its deployment approval process.
- What production-ready means for them.
- The integrations involved.
After we hand over
Who owns it after we leave
The named client engineers who went through the knowledge transfer. What it displaces: maintaining the system and the evaluation set — an estimate of the weekly time needs measuring during weeks 5–8, not guessing now.
What we measure
- One system metric agreed in week 1 (for example accuracy or cost per call)
- The number of client engineers who shipped a change on their own
What comes next
- An Embedded AI Engineerto keep momentum
- Managed Servicesto run the system
- A second FDE Sprint for the next system.
Questions buyers ask us
Yes — that is the AI Feature Sprint, and it is cheaper. This is for teams who need to own the system after we leave.
Fair. Then this is not the right programme, and we would rather tell you now than discover it in week four.
It ends in production, not in a demo. There is a working version in staging by week four.
We agree what production-ready means in week one. If something has to give, it is scope, not quality, and we say so at the time.
Mostly it is one engineer versus two, which depends on the system’s complexity and how many of your engineers we are bringing up to speed.