dev_surya

Junior Full Stack Dev 💻 | Remote 🏠 | I build real-world apps, share dev stories & crash hackathons for fun 🚀 | 22 🙆🏻

[dev_surya @suryaxdeveloper]

Back on Feb 5, 2025, our team Seed of Code competed in the Nosu AI Hackathon 2024 — a global AI innovation challenge with $11,300+ in prizes.

We ended up winning 🏆 Education & AI (3rd Place).
And this was just our second hackathon win. 😎
#100daysofcode #Hackathon #Winner

[dev_surya @suryaxdeveloper]

@vanshxdevs Yeahh bruhh

[dev_surya @suryaxdeveloper]

The more I scroll through X, the more I start questioning my skills. 😭

[dev_surya @suryaxdeveloper]

Going through Caching 0 → 1 🚀

Learned about:
• What caching is & why we need it
• Where to add caching in a system
• Caching architectures
• Cache eviction policies
• Common caching issues & their solutions

Also made a complete Notion note for revision before interviews. 📝

The more I learn System Design, the more fun it gets.

Understanding how systems actually work behind the scenes is genuinely exciting.

Now comes the interesting part - figuring out where I can actually apply these concepts in my projects. 👀

[dev_surya @suryaxdeveloper]

@vanshxdevs The problem is not only my problem anymore....

[dev_surya @suryaxdeveloper]

Good Morning 🌞 Sher aj jaldi uth gaya, ghar Jana hain naa isliye 😂

[dev_surya @suryaxdeveloper]

The last 6 months have humbled me quite a bit.

Throughout college, I focused heavily on development. I built projects, learned system design, won hackathons, worked with production-grade applications, became the Best Intern, and eventually worked on backend systems for OTT apps serving real users.

I also studied DSA up to Trees, but honestly, I didn't revise enough. I can solve easy/medium problems when I get enough time, but interviews don't always give you that time.

Then came the reality check.

• Rejected from TCS Prime after a ~5-minute interview and 3 questions.
• Rejected from Infosys today.
• Rejection emails from startups and remote startups almost every day.

The weird part?

When I left my internship in October 2025, I felt confident. Within two months, I cracked a remote Junior Full Stack Developer role.

After working for almost 9 months, handling backend applications, optimizing products, and working on production-grade OTT applications, I thought I'd naturally become more confident.

Instead, lately, I've been feeling the opposite.

Maybe it's the constant comparison on Twitter/LinkedIn.

Maybe it's interview anxiety.

Maybe my DSA preparation isn't strong enough.

Maybe I haven't learned how to communicate my experience properly.

Probably a combination of all of them.

But I'm realizing something important:

Being good at development ≠ being good at interviews.

And getting rejected ≠ being a bad developer.

So I'm going back to the basics.

DSA. Fundamentals. System design. CS concepts. Communication. Interview practice.

Less comparison.
More preparation.
More building.

The goal isn't to prove the internet wrong.

It's to become genuinely better than I was yesterday.

If you've gone through a similar phase, I'd genuinely love to hear what helped you get out of it.

[dev_surya @suryaxdeveloper]

@TheSuranaverse That's crazy 😂

[dev_surya @suryaxdeveloper]

Which tech stack is never going to die?

[dev_surya @suryaxdeveloper]

@TheSuranaverse I see!! BTW any reason??

[dev_surya @suryaxdeveloper]

@uday_devops Nest.js

[dev_surya @suryaxdeveloper]

@riya_mishra007 PostgreSQL

[dev_surya @suryaxdeveloper]

Mahesh Dalle, follow karle... 😖

[dev_surya @suryaxdeveloper]

Working on a quick commerce startup of my friend for our local area; do not know whether it will work or not.

Wish them luck!!!

[dev_surya @suryaxdeveloper]

Hey @X Family!

I’m Surya Ghosh — currently working as a Junior Full Stack Developer at AscentSpark Software, Kolkata (Remote).

I’m primarily focused on backend and full-stack development, working with NestJS, TypeScript, Node.js, PostgreSQL, AWS, Next.js, React.js and React Native.

At work, I’ve been involved in building and maintaining video-streaming platforms serving 200K+ users and 5M+ Play Store downloads, which has given me hands-on experience with authentication, APIs, databases, cloud infrastructure, video processing, streaming and backend systems.

Outside work, I’m deeply interested in backend engineering, system design, distributed systems and AI/LLMs.

One of my current side projects is SeedOfCode, an AI-powered personalized course generation platform. Users provide their learning requirements, skill level, available time and topics, and the platform generates a customized learning path along with relevant learning resources.

Recently, I’ve been going deeper into LLM engineering and AI infrastructure by building the system myself. I deployed a local Gemma model using Ollama on an Oracle Cloud VM, built a NestJS server around it, and implemented things like:

• Redis & BullMQ for queuing and asynchronous processing
• Worker Threads for concurrent processing
• API key management and rate limiting
• Self-hosted LLM inference
• AI-powered content generation and validation
• YouTube search and resource selection

I’m also exploring how to evolve SeedOfCode into a proper AI-powered learning platform, including AI-generated video explanations instead of relying entirely on external videos.

My current learning path is mainly:

Backend Engineering → System Design → Distributed Systems → AI/LLM Engineering

I’m trying to go beyond simply learning frameworks and understand how real-world systems are designed, deployed and scaled.

I enjoy building things from scratch, experimenting with new technologies, learning by doing, and turning ideas into working products.

GitHub: https://t.co/sFroCIJFYL
LinkedIn: https://t.co/enAlVViBbI
X: https://t.co/HIgLvwo9va
Email: [email protected]

[dev_surya @suryaxdeveloper]

I was using the Gemini API for one of my projects, but the rate limits started becoming a problem.

So I thought… why not self-host an LLM? 👀

Installed Ollama → downloaded a model → set up an Oracle Cloud VM → deployed it → built a server with NestJS → integrated it into my project.

Everything was going great until I realized…

Oracle VM2 has no GPU. 🥺

Now a simple question takes almost 1 minute to respond. 💀

Any better ways to deploy a local LLM or optimize inference for faster responses?

Would love to hear your suggestions! 👇

[dev_surya @suryaxdeveloper]

@phoenix_ykp As I’m currently working at a video streaming company, I’ve always dreamed of working at @netflix.

Apart from that, I’m open to any remote company, mainly @YCombinator-backed startups that pays well and offers opportunities to work onsite too.

[dev_surya @suryaxdeveloper]

@larrybson21 @Raynerdtech @imthepk @Khlaseek_dsgner @RoseMarvelous4 @Dev_JesseMaduka @FlutterDev Congratulations Man 🎉