Smit Waikar

Tech Innovator | AI Content Creator | Robotics Engineer | SEO Expert | Founder Orange Glazz | Where AI Meets Digital Excellence

[Smit Waikar @Smit_waikar]

Excited to share another step in my AI learning journey!
I’m happy to have earned the Project Management Assessment Certificate from LearnTube AI, powered by Google AI.
As I continue building my AI Project Management skills,
#AIProjectManagement #ArtificialIntelligence #PMP

[Brian @Inspector_9]

•◽️• @jn3008 + @etiennejcb

[Smit Waikar @Smit_waikar]

We are also in the race #swypee #orangeglazzai #KemalKılıçdaroğlu #spotify

[Smit Waikar @Smit_waikar]

@Spotify

[Smit Waikar @Smit_waikar]

@PeterDiamandis Link ?

[Smit Waikar @Smit_waikar]

@sujaldkl Can you please guild me

[Smit Waikar @Smit_waikar]

@weseeventures I Am interested as I am Ai Content Developer working now
I am fit candidate for this role with Content knowledge from 5 years

[Smit Waikar @Smit_waikar]

@tranmautritam For sure Ai can do it but need to be trained the motion graphics, key framing and all via prompt but prompting for such results will be the tuff part for sure

[Coder girl 👩‍💻 @dev_maims]

POV: Software engineers 3 hours before the deadline. 😂

[Smit Waikar @Smit_waikar]

MVP is the most important stage and we are heading towards it
Building Ai Powered company in 100 Days is not that easy but we are confident
#orangeglazz #swypee

[Shrijit More @ShrijitM_26]

Launching your MVP this week? Read this before you regret it. 🧵

Most founders don’t fail at launch…
They fail 48 hours after launch.
Here’s the exact 6-step pre-launch checklist we’re using before shipping our app:

- Observability (Don’t fly blind)
If users hit bugs before you see them → you lose.

→ Setup Sentry (errors)
→ Setup PostHog (user behavior)
- Security (People WILL try to break your app)
→ Validate every input (NoSQL injection is real)
→ Add rate limiting (bots are faster than you)
→ Lock CORS properly

- Auth Edge Cases (Silent killers)
This is where apps randomly break.
→ JWT expiry
→ Refresh tokens
→ Google/Auth data mapping

- App Store Reality Check ⚖️
If you're building a social/dating app:
Apple/Google WILL reject you without:
→ Delete Account
→ Block User
→ Report System
(No exceptions.)
- Stress Test BEFORE users do
Your API isn’t ready just because it works locally.

→ Simulate load (k6 / Artillery)
→ Test 500+ concurrent users
→ Watch DB choke in real time
- Legal + SEO (Most people ignore this)
→ Privacy Policy + Terms = mandatory
→ Setup Google Search Console
→ Start ranking from Day 1

We’re shipping our MVP this week.
This checklist probably saved us weeks of chaos.
Bookmark this. You’ll need it. 🚀
#buildinpublic #startup #webdev #indiehackers

[Smit Waikar @Smit_waikar]

@ridimabuilds We are building Ai Pet Dating App and need your help here is our logo

[Shrijit More @ShrijitM_26]

The 3D website era is here. And most people haven't noticed yet.

Spline just launched Omma — you can now build 3D websites and apps using AI agents. No big team. No months of work.

We spent years saying "3D on the web is too complex." That excuse is running out fast.

https://t.co/BgrmodlLqe

Credit: Spline (@splinetool)

[Shrijit More @ShrijitM_26]

Swypee Update: Dealing with 'The Pack' 🐕🐕

One bit of feedback I kept hearing: 'I have 3 dogs, I don't want to swipe 3 times.'

Today I shipped Batch Swiping. 🚀

Now you can select your whole pack and swipe once. Behind the scenes, it’s a NestJS batch endpoint wrapped in MongoDB transactions to keep things ACID-compliant. 🛡️

Matching just got 3x faster for multi-pet owners.

#BuildInPublic #ReactNative #NestJS #MongoDB #PetTech #Solopreneur

[Smit Waikar @Smit_waikar]

Congratulations Team India | Swypee is celebrating the journey of Win 🙌 #t20 #india #teamindia #viratkohli #msdhoni Thala for a Reason #RohitSharma #hardikpandya #Iran #swypee #swypeeapp #swypeeai #orangeglazz #indiasfirstaipoweredpetapp #WorldCup2026 #thalaforareason

[Shrijit More @ShrijitM_26]

The interviewer asked me these 3 questions about Agentic AI something every AI engineer should know.

Questions:

1️⃣ How do you handle memory bloating in agent systems?
2️⃣ How do you deal with partial failures in an agent workflow?
3️⃣ What performance issues occur in Agentic AI and how do you solve them?

⸻

Answers:

1️⃣ Memory Bloating

Agent systems store conversation history, reasoning traces, and tool outputs.
Over time this increases token size → slower responses and higher cost.

Ways to handle it:
• Context summarization for older interactions
• Store long-term memory in a vector database and retrieve only relevant data
• Token budgeting & pruning to keep the prompt small

⸻

2️⃣ Handling Partial Failures

Agent workflows depend on APIs, tools, and models.
Any component failure can break the workflow.

Good systems handle this with:
• Retry with exponential backoff
• Tool schema validation before execution
• Fallback models/services
• Circuit breaker patterns to isolate failing components

⸻

3️⃣ Performance Issues

Common bottlenecks in Agentic AI:

• Too many sequential LLM calls
• Large prompts causing token explosion
• Unnecessary tool usage

How to optimize:
• Parallel / async tool execution
• Use retrieval instead of full context
• Hybrid models (small models for simple tasks, larger models for reasoning)

⸻

Agentic AI isn’t just prompt engineering.
It’s about building reliable, scalable AI systems. 🚀

[Smit Waikar @Smit_waikar]

@playboiyng @CloverboysLIVE Lol MOMENTS

[Smit Waikar @Smit_waikar]

@mentxyk @CloverboysLIVE Why

[Smit Waikar @Smit_waikar]

@CloverboysLIVE Banger

[Smit Waikar @Smit_waikar]

@orange_glazz What's in your idea what can we add more into this

[Orange Glazz @orange_glazz]

Day 7