01 / LockedIn FDE Academy
Train for the whole production problem.
A connected development standard spanning field discovery, engineering, evaluation, governance, operating judgment, and transfer.
Explore the AcademyCareers / Forward-deployed engineering
Build AI where the work is real.
Forward-deployed engineers work inside the operation—mapping decisions, building the system, proving its behavior, and transferring it with evidence.
Join the talent network How we hire October 17: meetup at the labExplore the work
This small exercise makes one engineering decision tangible: a valid output still needs the right authority to act.
See how the work connectsField exercise / 001
Local workspaceEngineer the approval boundary.
Edit the policy, run the cases, and inspect what changes. Try setting requireHumanApproval to false to expose a release failure.
review-policy.json editable
The source list, confidence threshold and approval rule control the result. The task name labels your evaluation record.
Ready to evaluate
02 / Inspect a test caseSynthetic fixtures
A documented operating exception
An operator requests a review using the approved runbook.
- Source
- operations-runbook
- Fixture confidence
- 0.93
Expected behaviorPrepare the draft and hold it for a human decision.
03 / Evaluation record
Run the cases to create a recordThe evaluator compares the policy’s decisions with three fixed expectations. Editing the specification clears its result and any local approval.
Anatomy of the work
The work comes before the title.
Forward-deployed engineering connects technical depth to operating judgment. The same line of accountability runs from the first field note to the release record and the team that will own the system.
- 01
Enter the operation
Find the decision behind the request.
Work with operators to map the workflow, the source of truth, the authority boundary, and the evidence that would make a release acceptable.
- workflow map
- authority map
- acceptance brief
- 02
Engineer + evaluate
Build against the real boundary.
Connect models to approved context and scoped tools, then test the behavior that matters—including the actions the system must refuse.
- tool contract
- evaluation set
- trace record
- 03
Govern + release + transfer
Leave the operation able to own it.
Hold human authority where it belongs, make promotion reversible, and transfer the runbook, evidence, and operating knowledge with the system.
- approval record
- rollback path
- transfer runbook
Talent disciplines
One production problem. Many forms of depth.
The practice joins engineering, product, integration, transformation, trust, and client leadership around the same operating outcome.
These talent families describe the disciplines we build around. Formal role briefs name the jurisdiction, work pattern, and application path.
01/07Forward-deployed engineering
Own discovery, architecture, integration, build, release, and transfer inside the operating environment.
Representative paths
- Forward-Deployed Engineer
- Senior Forward-Deployed Engineer
- Principal Forward-Deployed Engineer
- Forward-Deployed Engineering Lead
02/07Applied AI and agent systems
Engineer agent runtimes, context systems, evaluation programs, and model behavior against real enterprise work.
Representative paths
- Applied AI Engineer
- Agent Systems Engineer
- Evaluation Engineer
- Machine Learning Systems Engineer
03/07Product and platform engineering
Turn field patterns into durable products, delivery accelerators, and production-grade platform capabilities.
Representative paths
- AI Product Engineer
- Full-Stack Product Engineer
- Platform Engineer
- AI Product Lead
04/07Data, integration, and cloud
Connect AI to the identity, data, APIs, event flows, and infrastructure that already carry the operation.
Representative paths
- Data Engineer
- Integration Engineer
- Cloud and Platform Engineer
- Enterprise Architect
05/07AI transformation and program leadership
Translate enterprise authority into operating models, portfolios, delivery systems, adoption, and measurable outcomes.
Representative paths
- AI Transformation Lead
- Technical Program Lead
- AI Operating Model Lead
- Adoption and Value Lead
06/07Trust, security, and compliance
Make authorization, security boundaries, control evidence, model governance, and operational assurance part of the build.
Representative paths
- AI Security Engineer
- Compliance Engineer
- AI Governance Engineer
- Control Evidence Engineer
07/07Client and partner leadership
Shape executable pursuits, align account and delivery authority, and carry the relationship through production acceptance.
Representative paths
- Engagement Architect
- Technical Deployment Lead
- Partner Delivery Lead
- Client Partner
Development + selection
Develop the judgment. Make the evidence legible.
The Academy and interview process share one standard: understand the operating context, make the boundary explicit, and show the decisions you carried.
02 / How we hire
Bring one piece of work into focus.
Conversations follow the context, decisions, tradeoffs, and evidence behind work you actually owned.
Read the interview guideStanding talent network
Show us the work you carry.
A strong introduction names the operating problem, the boundary, the decision you owned, and the evidence behind the result. Early-career builders can use a technical project, research effort, or product they carried deeply.