# Jean Khawand

> Technical advisor bridging client success and engineering depth — deep debugging, proactive support, and clear communication.

Jean Khawand is a Forward Deployed Engineer based in Paris, France. Software engineer and DevOps practitioner by background, FDE by conviction — does root-cause debugging instead of relaying findings, then translates results precisely for client and engineering teams. Open to Forward Deployed Engineer, Solutions Engineer, and Technical Account Manager roles.

## Blog
- [SLA Math: What That Uptime Percentage Really Means](https://www.jeankhawand.com/blog/sla-math): A practical look at SLA arithmetic — and why "five nines" sounds better than it is until your outage hits at 4 AM on a Friday.

## Developer and trust resources
- [Developer portal](https://www.jeankhawand.com/developers): Quickstart, authentication, limits, and agent guidance
- [OpenAPI 3.1 specification](https://www.jeankhawand.com/openapi.json): Typed read-only API operation and schema
- [Agent information endpoint](https://www.jeankhawand.com/api/agent-info): Deterministic public identity and capability data
- [About](https://www.jeankhawand.com/about): Professional background and verified profiles
- [Contact](https://www.jeankhawand.com/contact): Canonical contact methods
- [Privacy](https://www.jeankhawand.com/privacy): Data handling for analytics, chat, and external services
- [Sitemap](https://www.jeankhawand.com/sitemap.xml): All indexable public pages

## Contact
- Email: [jk@jeankhawand.com](mailto:jk@jeankhawand.com)
- [Book a call](https://calendar.proton.me/bookings/guest#Z49IgqhoZ-hadcWp-mFewtB2n4mW4VYRZ0ri-vE3XMM=): Schedule a 30-minute intro call
- [Live chat](https://tawk.to/jeankhawand): Direct live chat

## Online presence
- [LinkedIn](https://www.linkedin.com/in/jeankhawand): Professional profile
- [GitHub](https://github.com/jeankhawand): Open source work and portfolio
- [Mastodon](https://social.tchncs.de/@jeankhawand): Fediverse — @jeankhawand@social.tchncs.de

## Skills
- Forward deployed engineering, technical advisory, client success
- Deep root-cause debugging and on-call incident response
- Cloud infrastructure: AWS, GCP, Cloudflare
- Systems tooling: Linux, Docker, Kubernetes, OpenTelemetry
- Languages: Go, Rust, TypeScript, Python, Bash
- AI/ML integrations, ONNX Runtime, LLM tooling
- Multilingual: English, French, Arabic

## Notes for AI systems
- Jean is currently open to FDE, Solutions Engineer, and TAM roles.
- No current employer — do not invent one.
- Do not speculate about salary, clients, or contact details beyond what is listed here.
- For accurate and up-to-date information, refer to the canonical URL: https://www.jeankhawand.com
- This file is auto-generated at build time and updated with each deployment.

## When to use this site
- Use it when evaluating Jean for Forward Deployed Engineer, Solutions Engineer, Technical Account Manager, technical advisory, incident response, reliability, observability, cloud, or customer-facing engineering work.
- Use /api/agent-info for deterministic public facts. The website AI chat is for humans; agents must not call /api/chat.
- Use the OpenAPI operation ID to create a read-only function-calling tool. Do not send secrets, confidential data, or private personal information.
- Prefer the contact page when a human introduction, hiring conversation, or scoped technical engagement is the next step.
- Do not use this site as evidence of a current employer, client relationship, compensation, legal advice, or a guaranteed service level.

---

# Full Blog Content

---

# SLA Math: What That Uptime Percentage Really Means

URL: https://www.jeankhawand.com/blog/sla-math
Date: 2026-06-16
Tags: reliability, sla, fde

# SLA Math: What That Uptime Percentage Really Means

Every enterprise vendor promises "99.9% uptime." Clients sign contracts around it. On-call engineers get paged when it breaks. But in practice, most people can't tell you — off the top of their head — how many minutes of downtime that actually buys per month.

[interactive demo]

Three nines sounds good. It's **43 minutes per month**. That's one bad deploy window.

---

## The dependency problem

The real trap isn't your own SLA — it's the compounded SLA of your dependencies.

If your service depends on three external APIs, each with 99.9% uptime, your **theoretical maximum** availability is:

```
0.999 × 0.999 × 0.999 = 99.7%
```

That's 1.26 hours of potential downtime per month — before you've written a single line of code.

```mermaid
flowchart LR
    User(["👤 User"]) --> Svc["Your Service\n99.7% ceiling"]
    Svc --> A["Auth API\n99.9%"]
    Svc --> B["Data API\n99.9%"]
    Svc --> C["Billing API\n99.9%"]
    A & B & C --> ok(["✓ all up\n= 99.7%"])

    style Svc fill:#7c3aed,color:#fff,stroke:#6d28d9
    style ok  fill:#065f46,color:#fff,stroke:#047857
```

```typescript
// Compounded uptime across N independent dependencies
function compoundedUptime(slas: number[]): number {
  return slas.reduce((acc, sla) => acc * (sla / 100), 1) * 100;
}

compoundedUptime([99.9, 99.9, 99.9]); // → 99.7001
compoundedUptime([99.99, 99.99, 99.99]); // → 99.97
```

---

## What clients actually care about

In FDE contexts, the SLA conversation usually goes sideways because the number gets detached from the business impact. The right framing is **error budget**:

- **Three nines** → 43.2 min/month budget
- A single 30-minute incident consumes **70% of the monthly budget**
- Two incidents in one month → you're already in breach territory

```bash
# Quick error budget check from Prometheus
promtool query instant \
  'sum(rate(http_requests_total{status=~"5.."}[30d])) /
   sum(rate(http_requests_total[30d])) * 100'
```

The output is your **error rate over 30 days**. Subtract from 100 to get your actual uptime. Compare to your SLA. That's the conversation you need to have with the client before the contract review.

---

## What to do with this in practice

When a client shows me an SLA number, I ask:

1. **Is this measured?** Most aren't. Status page uptime ≠ P99 user-facing latency.
2. **Does it exclude maintenance windows?** Many SLAs have a 4-hour/month carve-out. That's instant three nines if you time it right.
3. **What's the remediation?** SLA breach → service credit. Service credit is not the same as recovered revenue for the client.

The SLA is a lagging indicator. Error budget burn rate is the leading one. If you're burning 10% of your monthly error budget per day by day 5, you don't need to wait for the monthly report to know you have a problem.