Inamul (@xInamul03) header
Inamul (@xInamul03) profile picture

LLM & AI Engineer | Building AI agents, RAG pipelines & local open-source stacks | Breaking down Claude updates & AI research papers into practical code.

Joined June 2025
Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)Photo by Inamul (@xInamul03)
Pinned Tweet
If you're into AI say hi Lets connect .... Follow <--> Follow back
26
Google is better positioned to replace Claude than OpenAI is....
1
24
Some Mfs doing masters to delay unemployment while Al keeps shrinking the job market...
31
Your RAG system gives wrong answers. Dumb AI engineer: 1.Changes the embedding model. 2.Increases top-K. 3.Adds a bigger LLM. 4.Rewrites the prompt. Smart AI engineer:
1
27
1.Evaluates retrieval quality. 2.Checks chunk boundaries. 3.Adds hybrid search. 4.Implements reranking. Same RAG pipeline. Different engineering maturity.
18
Inamul retweeted
KV caching is one of the biggest reasons LLMs generate tokens faster. Instead of recomputing attention for previous tokens, models reuse cached Key and Value states. Less redundant computation. Lower latency. Faster generation. A small optimization with a massive impact.
2
3
57
KV caching is one of the biggest reasons LLMs generate tokens faster. Instead of recomputing attention for previous tokens, models reuse cached Key and Value states. Less redundant computation. Lower latency. Faster generation. A small optimization with a massive impact.
2
3
57
Your AI agent keeps calling the same tool repeatedly. Dumb AI engineer: 1. Changes the system prompt 2. Adds another LLM call 3. Increases token limits 4. Switches to a bigger model 5. Blames the framework Smart AI engineer:
1
1
65
1. Inspects the execution trace 2. Checks tool outputs 3. Adds loop detection 4. Sets execution limits 5. Implements proper stop conditions Same agent. Different engineering maturity.
27
NVIDIA built a technology empire around Jensen Huang. Now, after admitting serious personal mistakes, he's leaving the door open to becoming CEO again. But should a company's future depend on one person, no matter how influential?
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation. Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
3
1
57
NVIDIA's success and Jensen's personal accountability are two separate questions. One shouldn't erase the other. The real challenge is ensuring that corporate leadership is built on both exceptional results and public trust.
3
A CEO's legacy shouldn't give him immunity from accountability. Jensen's contribution to NVIDIA is undeniable, but leadership also means accepting consequences. The Board now faces a difficult question: how should it balance past achievements with present responsibility?
5
The real test isn't whether Jensen Huang can return as CEO. It's whether NVIDIA's Board believes his leadership is still worth the risks. No company, no matter how successful, should be bigger than accountability.
1
16
OpenAI drops GPT-6.2. Meanwhile, developers are still debating whether GPT-6.1 is better than GPT-5.6. AI models are evolving faster than developers can test, compare, and adapt. At this point, we're not just racing against AI. We're racing against the release cycle. 😂
And here we go…GPT-6.2 Astra and GPT-6.2 Sol are yours to make more games with. Oh, and they’re also super efficient and really good at making slides, building websites, and tackling the work you were doing before you started making games. Rolling out today to ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu users. openai.com/index/introducing…
4
2
54
AI agents moving from chat windows into live, multi-participant video streams as active team members. The shift from synchronous prompt-response to asynchronous background agent teams is officially here.
2
Don't judge Fable 5.2 by its price tag or a few random prompts. If you get the chance to try it, give it a real challenge. Use it for deep research, complex reasoning, coding, and long-running tasks. The real value of an AI model isn't how impressive its demo looks.
3
My friend got rejected in an AI Engineer interview over this RAG question: Your vector DB holds outdated embeddings after millions of documents change. How would you keep retrieval fresh without re-embedding everything or increasing latency? How would you design it? Thoughts?
3
Damn I am using grokbot more than Claude code now. I’m a power user of Claude code burning 3-5B tokens a day.
13
Damn I am using grokbot more than Claude code now. I’m a power user of Claude code burning 3-5B tokens a day.
1
1
62
My friend got rejected in an AI Engineer interview over this RAG question: Your vector DB holds outdated embeddings after millions of documents change. How would you keep retrieval fresh without re-embedding everything or increasing latency? How would you design it? Thoughts?
Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.
5
1
2
94
AI agents moving from chat windows into live, multi-participant video streams as active team members. The shift from synchronous prompt-response to asynchronous background agent teams is officially here.
8