varun

19. Founding engineer backend & devsecops Engineer @Dynamic__Tech

[varun @VarunSheesh]

i'm varun, 19. backend & devSecOps Engineer at DynTech. building CDeX while studying CS full-time. I spend my time on backend internals, infra, distributed systems, and shipping under pressure. from not knowing loops to building real systems. Just gettin started.
https://t.co/v5nh6DESam

[varun @VarunSheesh]

@TimsBila didnt got one

[varun @VarunSheesh]

Everyone wants to be part of a startup.

Until they realize it means actually building one.

We’re 4 working professionals building a social app on the side, and we’re looking for one UI/UX designer who’s willing to collaborate, learn, and actually ship.

You don’t need to give us 5 hours a day.
If you can consistently spare even 30 minutes, that’s enough.
We’re not looking for someone chasing a fancy title or waiting for perfect conditions.
We’re looking for someone who’s curious, reliable, and wants to build alongside people who’ve already been shipping products professionally.

If that sounds like you, my DMs are open.
If not, keep scrolling.

[varun @VarunSheesh]

@classic_servers yepp
dm’s are open

[varun @VarunSheesh]

@vibewithfidelis dm’s are open

[varun @VarunSheesh]

@darijan___ @8xsocial https://t.co/6mPJblizvE

intrested ~

[varun @VarunSheesh]

@code_bytein true, its simple if u don’t face why the particular thing was slow and why you would upgrade to another methods

[dhyan.exe @dhyann_exe]

building a social app where our main kpi is how fast we can get you to close the app and go outside is kinda funny.

we’re onboarding our first users soon. if you want a spot, register for early access at https://t.co/L5yQJQMRR4

[varun @VarunSheesh]

@anurag__kochar with a vibecoded website 😭🙏

[varun @VarunSheesh]

peak's backkkkkk

[TVアニメ「ブラッククローバー」2nd Season【公式】 @bclover_PR]

━━━━━━━━━━━━━━━━
TVア��メ「#ブラッククローバー」
　　　　　2nd Season
　　　　　本PV公開⚔️
━━━━━━━━━━━━━━━━

迫力あるアクションと、
“絶望”が迫るダークな世界観を描いた本PVを公開！

2nd Seasonで描かれる激闘の一端を
ぜひご覧ください🔥

2026年10月から
テレ東系列ほかにて放送開始📺

#ブラクロ #BlackClover

[dhyan.exe @dhyann_exe]

We have thousands of "connections" and "followers" online, but when was the last time you met someone new in real life?

Every modern app is a trap designed to maximize your screen time. It’s making us isolated.

That's why we built Talphy.
A platform meant to get you off your phone, go check it out drop your email on the waitlist and secure early access

https://t.co/ICd3eoX7F8

[Yoshik @AskYoshik]

Stop doing DevOps in 2026 like it’s still 2019 “I installed Prometheus, I’m done” mode.

Do this instead.

1. Infra & environments
- Pick one app (Django, Node, whatever) and run full infra as code:
Terraform for VPC, subnets, RDS, S3, IAM, ALB.
Separate dev/stage/prod. Remote state, workspaces, basic policies.

- GitOps it:
ArgoCD or Flux syncing a k8s cluster, PR-based changes only.
Feature flags, blue/green or canary on the same app.

- Break it:
Rotate RDS password with zero downtime.
Kill a node, test pod disruption budgets and readiness checks.

2. Observability & SRE basics
- Full stack tracing/logs/metrics, not just “Grafana looks nice”:
OpenTelemetry everywhere. Central log pipeline with Loki/ELK.
RED/USE dashboards for one service.

- Error budget for a real endpoint:
Define SLOs on login or checkout.
Hook alertmanager / PagerDuty. Tweak alert noise based on incidents.

- Run a game day:
500s from a dependency, DNS failure, Kafka lag.
Do postmortems. Improve runbooks, dashboards, and alerts.

3. Platform & AI-in-prod glue
- Build a small internal “platform”:
Simple API for app teams: namespace per team, default network policies, autoscaling, shared postgres, S3 buckets.
Self-service templates using Helm or Backstage.

- Add an LLM service like any other workload:
vLLM or Ollama in k8s. GPU node pool. HPA on QPS or latency.
Token usage, cost per request, and rate limiting in the gateway.

- Wire it all together:
CI/CD with GitHub Actions/GitLab. Image signing, basic SBOM.
Canary deploys, quick rollback, feature flag switch.

Keep in mind for any DevOps project:
- Always have IaC, not click-ops.
- Put rate limiting, caching, and health checks in front of everything.
- Think SLOs first, dashboards second.
- At mid/senior, focus more on k8s, load balancing, multi-tenant clusters, and infra reliability than pretty project names.

Instead of paying $100 on random certs, burn that on cloud credits, RDS, EKS/GKE, and a cheap GPU box.

Use Claude/ChatGPT as your pair SRE: design docs, runbooks, incident drills. Ask annoying “what if this dies?” questions until the system is boring.

[dhyan.exe @dhyann_exe]

Guys waitlist's is open go check it out!!🚀

[varun @VarunSheesh]

@dodopayments I still prefer @Razorpay ✌🏻

[varun @VarunSheesh]

@amyy_tweetz

[varun @VarunSheesh]

ditch tinder and hinge, make irl connections instantly n meet irl.
waitlist’s open at https://t.co/vo32q4hFq8

[varun @VarunSheesh]

well you can go for AI engineering there’s a playlist by @sriniously backend from first principles, make notes, make projects on ur own applying the theory , then after you get a good hold in backend refer this https://t.co/k86BzrWIfa for AI engineering
(grind dsa lil bit side by side if possible as well)

[varun @VarunSheesh]

anyone out there who still writes code with their fingers :/

[miyae 𐙚 @ria_cheal]

My little escape

[varun @VarunSheesh]

starting the day wid claude max limit reached, lets see how it goes