Kard

13/03/2026

AI in 2026

21/02/2026

I'm coding for both oooh😩🤦😂

31/01/2026

I served as the assistant biomedical engineer during a successful brain surgery yesterday.
I had the honours of making sure that all the medical devices needed were in good working condition and that the procedure was without hitches 😁
It was a very sweet first time experience for me in a neurosurgical environment 😚
More to be done in this biomedical field
POV: I am also an AI/ML engineer 😁

07/01/2026

AI doesn't sit in meetings, neither can they fully understand the needs of a software. In as much as it has a mind of its own, it still works on the basis of GIGO and that is gotten from the prompt made available

07/01/2026

BREAKING BLISS: Bangladesh Powering Up the "Silicon River" at BRAC University!
A historic milestone for the nation's tech landscape was reached yesterday, December 29, 2025, with the official inauguration of the Centre of Research Excellence in Semiconductor Technology (CREST) at BRAC University.
This national hub marks Bangladesh’s official entry into the global semiconductor race, moving the country from a technology consumer to a high-impact innovator.
🔬 What is CREST?
As the intellectual nucleus of the "Silicon River" ecosystem, CREST will focus on:
• AI-Driven Chip Design & VLSI Systems
• Robotics & Next-Generation Hardware
• Advanced Semiconductor Packaging & Testing
• Silicon Photonics & Low-Power Devices
🚀 A Unified Vision: The BEAR Framework
The center was launched under the broader BEAR framework—integrating Biotech, Electronics, AI, and Robotics—to position Bangladesh as a regional deep-tech leader.
🎙️ Voices from the Launch
"Bangladesh’s future as a ‘Nation of Innovation’ will be defined by the global impact of its ideas, not just its population size." — Professor Muhammad Mustafa Hussain (Purdue University), Architect of the Silicon River vision.
Vice-Chancellor Professor Syed Ferhat Anwar reaffirmed BRAC University's role as the academic anchor for this national priority. Industry giants like Neural Semiconductor (Principal Sponsor), Synopsys, and GlobalFoundries have already pledged their support through global-standard tools and curriculum alignment.
🎓 Nurturing Talent
During the ceremony, nine graduate researchers from top universities (including BUET, DU, and NSU) were awarded the first CREST Fellowships to lead pioneering semiconductor research.
The future is silicon. The future is Bangladesh. ⚡

🌐 BREAKING BLISS: Bangladesh Powering Up the "Silicon River" at BRAC University!

A historic milestone for the nation's tech landscape was reached yesterday, December 29, 2025, with the official inauguration of the Centre of Research Excellence in Semiconductor Technology (CREST) at BRAC University.

This national hub marks Bangladesh’s official entry into the global semiconductor race, moving the country from a technology consumer to a high-impact innovator.

🔬 What is CREST?
As the intellectual nucleus of the "Silicon River" ecosystem, CREST will focus on:

• AI-Driven Chip Design & VLSI Systems

• Robotics & Next-Generation Hardware

• Advanced Semiconductor Packaging & Testing

• Silicon Photonics & Low-Power Devices

🚀 A Unified Vision: The BEAR Framework
The center was launched under the broader BEAR framework—integrating Biotech, Electronics, AI, and Robotics—to position Bangladesh as a regional deep-tech leader.

🎙️ Voices from the Launch
"Bangladesh’s future as a ‘Nation of Innovation’ will be defined by the global impact of its ideas, not just its population size." — Professor Muhammad Mustafa Hussain (Purdue University), Architect of the Silicon River vision.

Vice-Chancellor Professor Syed Ferhat Anwar reaffirmed BRAC University's role as the academic anchor for this national priority. Industry giants like Neural Semiconductor (Principal Sponsor), Synopsys, and GlobalFoundries have already pledged their support through global-standard tools and curriculum alignment.

🎓 Nurturing Talent
During the ceremony, nine graduate researchers from top universities (including BUET, DU, and NSU) were awarded the first CREST Fellowships to lead pioneering semiconductor research.

The future is silicon. The future is Bangladesh. ⚡

27/12/2025

Johannes Kepler was a visionary astronomer who transformed our understanding of the cosmos, giving us the three fundamental laws of planetary motion. His contributions are foundational to modern astronomy and helped lay the groundwork for other pioneering discoveries, such as Sir Isaac Newton’s theory of gravity.

27/12/2025

Have you ever wondered how your phone recognizes your face, how Google Photos knows a dog from a cat, or how voice assistants understand what you’re saying? Behind all of that is a powerful technology called neural networks — and this chart gives us a simple way to understand how they work.

Imagine showing a computer a photo of a dog. At first, it sees just pixels — tiny dots of color. But as the image moves through different layers of the neural network, the system starts to recognize edges, shapes, and patterns. Eventually, it understands that those features add up to something familiar: a dog.

Each layer in the network has a job. The first layer detects basic things like brightness and color. The next layers find edges, combine those edges into shapes, and then identify features like ears or fur. By the final layer, the system can confidently say, “That’s a dog.”

This process is what powers image recognition, voice assistants, recommendation systems, and even medical tools that help doctors spot diseases. Neural networks don’t just memorize — they learn. And they can get better over time.

The chart also shows two types of networks: shallow and deep. Shallow networks have fewer layers but more neurons per layer — they’re powerful but can be slow and complex. Deep networks have many layers with fewer neurons, which makes them better at understanding abstract ideas and solving harder problems with less effort.

🌍 Why This Matters to You
Neural networks are already part of your daily life. They help:

Your phone unlock with your face

Apps translate languages instantly

Online stores recommend products you’ll actually like

Doctors detect cancer earlier using smart scans

Banks spot fraud before it affects you

This technology is quietly working behind the scenes to make life easier, safer, and more personalized. And as it continues to evolve, it will help us solve even bigger challenges — from climate prediction to smarter education and healthcare.

So next time your phone recognizes a photo or your playlist feels “just right,” remember: it’s not magic. It’s a neural network doing its job (layer by layer, neuron by neuron).

21/12/2025

Medical imaging has always been central to cancer diagnosis, using tools such as X‑rays, CT scans, MRIs, and mammograms to reveal abnormalities inside the body. Traditionally, radiologists examined these scans manually, searching for suspicious shapes, densities, or irregularities. Today, artificial intelligence has transformed this process by being integrated directly into imaging devices, allowing them to highlight potential cancerous regions automatically.

The journey begins with data acquisition, where the imaging device captures detailed pictures of tissues and organs. These images are then processed by machine learning models that have been trained on thousands or even millions of past cancer cases. By learning the subtle differences between healthy and abnormal tissue, the system can detect tumors that may be too small or too complex for the human eye to consistently identify.


Once the images are processed, AI systems apply techniques such as pattern recognition to identify shapes, textures, and densities that match known cancer signatures. They also perform segmentation, which involves outlining suspicious areas and separating them from surrounding tissue for closer analysis. Classification follows, where the system predicts whether the abnormality is likely benign or malignant, often providing a confidence score to guide doctors. In breast cancer screening, for example, AI can scan mammograms and flag subtle changes that radiologists might miss, improving early detection rates.

The impact of this technology is profound. Automated cancer detection improves accuracy by matching or even exceeding human radiologists in certain cases, especially in early stages. It speeds up the diagnostic process by reducing the time needed to analyze scans, and it ensures consistency since AI systems do not tire or vary in judgment. Importantly, these tools do not replace radiologists but act as a second set of eyes, supporting doctors in confirming findings and reducing errors.

Looking at the bigger picture, the evolution of automated cancer detection reflects decades of innovation. Early imaging provided static pictures, but now AI transforms those images into actionable insights. This progress is the result of countless experiments, breakthroughs in deep learning, and collaboration between engineers and clinicians. As algorithms continue to improve, medical imaging devices will not only detect cancer but also predict treatment outcomes, personalize therapies, and monitor progress in real time. The ultimate goal is clear: faster diagnoses, earlier interventions, and better patient survival rates.

21/12/2025

Neural networks sit at the very heart of modern artificial intelligence, and this chart highlights the incredible diversity of forms they can take. Each type of network has its own unique way of learning, adapting, and solving problems. Some specialize in recognizing images, powering technologies like facial recognition and medical imaging. Others are built to understand language, enabling translation tools and conversational AI. Still others are designed to generate entirely new content, from realistic images to original text and music.

What makes this ecosystem so fascinating is how each model builds upon the work of those that came before. The earliest, simplest networks laid the foundation decades ago, proving that machines could learn patterns from data. Over time, researchers introduced more advanced architectures capable of handling memory, processing sequences, and even simulating creativity. These innovations didn’t happen in isolation they represent a continuous chain of breakthroughs, each one opening doors to new possibilities.

Today, neural networks form the backbone of the AI systems we interact with daily. They drive recommendation engines that suggest what to watch or buy, power voice assistants that understand our commands, and enable tools that can analyze massive datasets in seconds. Looking at this chart is more than just a technical overview it’s a reminder of the remarkable journey of innovation. Progress in AI has never been instant; it’s the result of countless experiments, bold ideas, and persistent refinement.

And the story is far from over. As new types of networks continue to emerge, the boundaries of what AI can achieve will keep expanding from smarter healthcare solutions to more personalized education, and even creative collaborations between humans and machines. The future of neural networks isn’t just about technology; it’s about shaping how we live, work, and imagine what’s possible.

17/12/2025

Every day, millions of posts are shared across platforms like Twitter, Facebook, and Instagram. Some are lighthearted, others are heated debates, and unfortunately, a portion cross the line into toxic or harmful language. When this happens, accounts often get flagged, sometimes leaving users confused about why it occurred. The truth is that platforms cannot rely solely on human moderators to sift through this endless stream of content. Instead, they turn to artificial intelligence models trained on massive datasets of toxic versus non‑toxic posts.

I recently trained a machine learning model using a dataset from Kaggle that had been previously classified by thousands of people into toxic and non‑toxic categories. This kind of crowdsourced labeling is powerful because it captures the collective judgment of diverse users, and it gives AI systems a foundation to learn what harmful language looks like in practice. By cleaning the tweets, converting them into numerical features, and experimenting with different algorithms, I was able to see firsthand how models can distinguish between safe and toxic content. Random Forest, in particular, stood out for its balanced performance, while Logistic Regression showed strength in catching more toxic tweets, even if it sometimes flagged safe ones.

These models learn patterns in language — certain words, phrases, or combinations that often signal harassment, hate speech, or abuse. In my own experiment analyzing toxic tweets, I discovered that while non‑toxic content dominates (over 90%), AI models are able to identify toxic language with varying degrees of success. Logistic Regression, for example, was better at catching toxic tweets, but sometimes flagged safe ones too. Random Forest struck a balance, achieving strong accuracy while making fewer mistakes overall. This mirrors what happens on real platforms: some accounts are flagged even when the intent wasn’t harmful, because the AI errs on the side of caution, while other accounts slip through because toxic language can be subtle, sarcastic, or coded.

The important point is that AI is not perfect, but it is essential. Without it, harmful content would overwhelm moderators and communities. With it, platforms can filter the majority of toxic posts, leaving human reviewers to handle the edge cases. As social media continues to shape conversations globally, understanding how and why accounts are flagged helps us see the bigger picture. AI isn’t silencing voices — it’s protecting communities, creating safer spaces for dialogue, and ensuring that the digital town square doesn’t become a hostile environment.

17/12/2025

Data science isn’t just about crunching numbers; it’s about understanding purpose, people, and process. At its core, it provides a framework that helps us see beyond the data itself and into the decisions it drives.

This framework can be thought of in six dimensions. It begins with why we do it — the goals and problems we aim to solve. Then comes how we do it, the methods and approaches that guide our analysis. We also consider who’s involved, because collaboration across teams and disciplines is essential. From there, we look at what steps we take, the structured process that turns raw information into insights. Alongside this, we examine what tools we use, the technologies and platforms that make the work possible. Finally, we recognize how culture shapes it, since the values and environment of an organization influence how data science is applied.

Whether you’re in tech or not, this perspective shows how data science fits into real‑world decision‑making. It’s not just about data; it’s about using information with clarity and purpose to create meaningful impact.