Joined January 2026
68 Photos and videos
Python Philosophy
2
9
Saad Khan retweeted
Jun 7
I made an app to track my Claude and Codex token usage in my Mac notch menu bar. Shows the exact refill times and everything. Try it out below :)
1
1
3
95
To become a strong Devops engineer dont just learn tools Understand how each tool supports business goals
1
13
Saad Khan retweeted
Jun 4
50% off for 10 of you: 50FOR10
1
4
115
Saad Khan retweeted
Jun 3
I made using Claude and Codex super easy with token usage, servers, paths, and much more all in my Mac notch menu bar. Try it out below :)
3
1
18
2,755
Saad Khan retweeted

3
4
23
8,819
Vertical Scaling in GCP (Scale Up/Down) Simple rule: Stop the VM first. Persistent Data stays safe. SSH keys remain intact. Local SSDs? Force stop = Data loss. Sustained Use Discount? Resets. From n1-standard-1 to e2-custom-4-2048 in under a minute. #GCP #CloudAdmin
1
17
Never overpay for unused vCPU or RAM again! The "Custom Machine Type" in #ComputeEngine is real magic.Need 6 vCPUs and 21.5 GB RAM? Done. (In other clouds, you'd pay for the next tier up).Perfect for cost-optimized architecture. #GoogleCloud #FinOps #Architecture
2
62
Saad Khan retweeted
May 31
Use Claude and Codex in your Mac notch and menu bar. New landing page, check it out below :)
2
1
10
1,408
GCP's bestkept secret for beginners? When you create a VM on the console, scroll DOWN. Click "Equivalent Command Line." Google writes the ENTIRE gcloud script for you. No need to memorize syntax. Copy, paste, and automate. This is a game-changer. #GCP #DevOps #CloudEngineer
2
28
Saad Khan retweeted
i still think about this
210
461
18,161
1,059,413
The "Budget Airline" of Cloud Computing ✈️: GCP Preemptible VMs. ☑️ Up to 91% cheaper than regular VMs ☑️ Perfect for batch jobs & rendering ⚠️ Can be "evicted" anytime (24h max life) ⚠️ No SLA, no guarantees. Not for prod databases! Cloud economics at its peak! 💸 #GCP #FinOps
1
26
REST vs gRPC in K8s - HTTP/1.1 REST = Head-of-Line blocking, JSON parsing tax, connection overhead - HTTP/2 gRPC = True multiplexing, binary protoBuf (5x smaller, 10x less CPU), 4 streaming patterns K8s control plane already uses gRPC natively!
1
1
19
Will GenAI kill Kubernetes? NO. Period. Why K8s isn't going anywhere: - $200 BILLION invested in CNCF ecosystem - Security tools built around K8s - Monitoring stack Prometheus/Grafana - Service Mesh (Istio, Linkerd) - 5M K8s developers globally AI will USE K8s. Not REPLACE it
1
24
Why Google Cloud is the speed demon Per-Second Billing (1 min min): First cloud to do it. Pay only for what you run. KVM Hypervisor: Open-source, blazing fast boot times (~30s). Default Encryption: Secure by design, no extra checkbox needed.
1
1
42
Same hardware powering YouTube & Gmail is powering your VMs.
1
13
GCP's "Secret Sauce" for VMs: CUSTOM MACHINE TYPES AWS says: "Pick 4vCPU/16GB RAM." GCP says: "I need 6vCPU/13.25GB RAM exactly." No overpaying for unused resources. But remember: vCPU: Even numbers only (2,4,6,8) RAM: Multiples of 0.25 GB only. Perfectly tailored #Compute
1
20
K8s OOM Killer: Who Dies First? QoS Classes (Order of Sacrifice): 🥇 BestEffort – No requests/limits → FIRST to die 🥈 Burstable – Requests < Limits → SECOND to die 🥉 Guaranteed – Requests = Limits → LAST to die The Problem: No requests set → Scheduler
3
1
48
Fix: Set requests = limits Use LimitRange Reserve memory for OS (--system-reserved)
1
11