RitikaxG

Backend & Applied AI Engineer
Building reliable AI workflows, agents, RAG, memory, evals & infra.
Looking to work with early-stage AI/startup teams.

[RitikaxG @RitikaxG]

Built SpinUp — a small Replit/Bolt-style control plane that turns a project idea into a live cloud workspace.

Github Repo : https://t.co/aQfYdciTtY

[RitikaxG @RitikaxG]

@iamdhananjayb @X Happy to connect

[RitikaxG @RitikaxG]

@Vvikramai Hi Vikram, looking forward to connect with you

[RitikaxG @RitikaxG]

@sharma__jii__ Happy to connect with you

[RitikaxG @RitikaxG]

@SamyaakJain @X Hello Samyaak, happy to connect

[RitikaxG @RitikaxG]

@SAMK_online Happy to connect

[RitikaxG @RitikaxG]

@ThisIsBhandari That’s amazing. Looking back it’ll all be worth it

[RitikaxG @RitikaxG]

@akashrajsarraf @X HI Akash, happy to connect

[RitikaxG @RitikaxG]

@DanielSmidstrup @X Hi Daniel, happy to connect

[RitikaxG @RitikaxG]

@GohilHardy Building workflow automation system for insurance claims. Happy to connect

[RitikaxG @RitikaxG]

@AnandButani Hi Anand, happy to connect

[RitikaxG @RitikaxG]

@buildwithmaya Hi Maya, I am building agentic workflow automation system for insurance claims. Happy to connect

[RitikaxG @RitikaxG]

@sharma__jii__ Happy to connect

[RitikaxG @RitikaxG]

@skar_connect Happy to connect

[RitikaxG @RitikaxG]

@joseph_mugen Hi Mugen , working in ai automation. Happy to connect

[RitikaxG @RitikaxG]

@dailykebe Hi Kerstin, happy to connect

[RitikaxG @RitikaxG]

@RachitThakur146 @X Hi Rachit, happy to connect

[RitikaxG @RitikaxG]

@lucas_builds Fellow developer here, happy to connect

[RitikaxG @RitikaxG]

What actually goes behind running an async task in JS?

Beginner mistake:

const result = await apiCall();
…and assuming everything will work smoothly.
Production-style async work is different.

Each async task may run concurrently, but it should be monitored separately for success, failure, retries, timeout, latency, and errors.

1. Use Promise.all() when all tasks must succeed.

2. Use Promise.allSettled() when tasks are independent and every result matters.

3. Wrap async boundaries with try/catch so you know why something failed.

4. Use typed errors so the backend can decide:

Should we retry?
Should we fail permanently?

5. Every external request should have a timeout. Use cancellation to stop useless underlying work and save resources.

6. Differentiate retryable failures like network/API issues from non-retryable failures like invalid JSON or COST_LIMIT_EXCEEDED.

7. Use backoff and jitter while retrying. Workers should not retry immediately and overload the backend again.

8. Log status, latency, error type, retry count, and retryability.

That’s what makes an async task production-ready.

[RitikaxG @RitikaxG]

How do design choices affect backend performance drastically.
Lets take an example of Uptime Monitoring Platform

Bad Design
500 websites fetched -> Pushed all a once -> Leads to redis network spike -> workers gets flooded -> event loop gets delayed -> downstream systems becomes unstable

Backend = Overloaded

Good Design
500 websites fetched -> Pushed in controlled batches -> Redis receives predicted load -> Worker consumes continually -> downstream systems remain stable

Backend = Happy happy

[RitikaxG @RitikaxG]

@joshjfish @X Hi Josh, happy to connect