Selected work

Decisions, constraints, and results from work inside live systems.

Each preview keeps the situation, one consequential decision, the visible result, and the related problem together. Client details stay anonymous where permission or an NDA requires it.

What these cases disclose

Claims use the narrowest wording supported by the source record. Client names, extra metrics, and sensitive mechanics stay out unless they are explicitly cleared for publication.

Flagship cases

One flagship case for each problem domain.

The order matches the four operating situations on the homepage, so the route from problem to proof stays easy to follow.

  1. Feature Delivery Recovery

    Delivery Recovery

    Inherited platform recovery after a leadership handoff with no reliable system map.

    Decision
    Use a full boundary redesign instead of pretending local fixes would restore delivery.
    Result
    Delivery became more legible and predictable before any public metric is used.
    Role
    Chief Technology Officer
    Read the case
    Delivery Recovery
  2. System Digital Twin

    AI Engineering Control

    An inherited platform was too large and undocumented for manual reasoning under deadline.

    Decision
    Build a textual system twin that agents could navigate progressively.
    Result
    The team regained an inspectable map of the system without exposing gated metrics on public surfaces.
    Role
    Chief Technology Officer
    Read the case
    AI Engineering Control
  3. Redis P0 Before First Users

    AI Product Risk & Safety

    A pre-launch product safety audit treated session storage as a real attack surface.

    Decision
    Remove bearer tokens from Redis and add a deterministic gate that checks storage contents directly.
    Result
    The case stays on the storage boundary and remediation, without making broader product claims.
    Role
    Founding AI Product Architect
    Read the case
    AI Product Risk & Safety
  4. Layered DDoS Defense Beyond the CDN

    Resilience & Security

    A live platform needed protection against behavior-aware attack traffic beyond generic volume.

    Decision
    Keep Cloudflare on the volumetric layer and add a separate application-behavior layer behind it.
    Result
    The case keeps the two defense layers distinct and omits unapproved metrics.
    Role
    Chief Technology Officer
    Read the case
    Resilience & Security

Supporting cases

Supporting cases add the context around each intervention.

These cases cover the diagnosis, adoption pattern, or trade-off around the flagship result. They do not count the same metric twice.

  1. Delivery RecoverySupporting case

    First Day After the CTO Exit

    Delivery Recovery

    A new CTO inherited production pressure and no usable handover.

    Decision
    Map incidents and ownership before promising acceleration.
    Result
    The preview stays on diagnosis and operational clarity, not on duplicate metric counting.
    Role
    Chief Technology Officer
    Read the case
    Delivery Recovery
  2. AI Engineering ControlSupporting case

    Agent Library Adoption by Pull

    AI Engineering Control

    Formal AI enablement created compliance theatre instead of habit change.

    Decision
    Leave a useful agent library in the shared repo and let adoption happen by pull.
    Result
    The preview records an observed adoption pattern without productivity or superiority claims.
    Role
    Chief Technology Officer
    Read the case
    AI Engineering Control
  3. AI Product Risk & SafetySupporting case

    Bonus Abuse as an Account Factory

    AI Product Risk & Safety

    Fraud recurred through linked account creation rather than isolated bad actors.

    Decision
    Treat the operation as a factory graph instead of a queue of separate accounts.
    Result
    The case stays non-numeric and withholds the detection mechanics.
    Role
    Chief Technology Officer
    Read the case
    AI Product Risk & Safety
  4. Resilience & SecuritySupporting case

    GCP to Bare Metal: Cost and Control

    Resilience & Security

    Cloud cost and network-control limits became part of the same operational problem.

    Decision
    Move the platform to infrastructure where the team could own the critical network boundary.
    Result
    The preview keeps the trade-off visible without flattening the case into anti-cloud rhetoric.
    Role
    Chief Technology Officer
    Read the case
    Resilience & Security

Related problems

Each case still belongs to a buyer situation.

Use the problem routes to inspect the trigger, stakes, and fit boundaries behind each case family.

Next step

Facts and context stay more useful than a polished brief.

Facts keeps the stable public reference. Start is the neutral route when the situation is visible but not yet framed as an engagement.