Portfolio Landing page

Who Am I ?

About Me

Hey, this is Sakib — a tech enthusiast — and I have a huge passion for building cool stuff. My main interests lie in Machine Learning and Data Science, but I’m also well-versed in cross-platform app development (React Native), native app development using Android Studio (Java), web development as well as backend development — creating scalable systems and APIs. I also explore frontier technologies such as Augmented Reality, Edge Computing, Game Development (Unity/C#), and STEM.
So far, I’ve fine-tuned numerous base LLMs like LLaMA 2 and Granite across various domains such as education, healthcare, astrophysics, quantum physics, code generation, task summarization, text generation, and more. Notably, I even designed my own language model architecture to reduce computational costs and memory usage.

Why Hire Me:

Because of my experience on multiple domain of technology {App development, Machiene Learning, Full Stack Development, Game Development} with real world project I can clearly imagine how different parts of a project connect and interact on big archtecture. This kind of full-stack understanding helps me visualize the entire workflow, avoid roadblocks, and collaborate smoothly with teams working on different parts of the system. Aditionaly I have real-world experience as I have been involved in many institutions and organizations, which improved my people skills, teamwork, and team management abilities. Perticipating in world-renowned hackathons and competitions like Microsoft Imagine Cup, and global finalist postion in my selve at the NASA Space Apps Challenge further enhanced my technical skills.

Download CV

What I Do ?

Service

AI Engineer/ Data Sciencet

Finetuned many base model and also created my own LM archtecture.

Cross Site App Developer

Developed Cross Platform app in React Native and native androind app in Android Studio Java

Unity Game Design

I HAVE VERY WELL VERESD WITH UNITY C# as i developed xr software like Vitro to accelrate xr education.

Backend Development

Labore velit culpa adipisci excepturi consequuntur itaque in nam molestias dolorem iste quod.

Full Stack Development

Mobile/Web App → Train ML Model → Deploy to Cloud → Create API → Integrate with App.

Cross Platform & Native App Developer

React Native & Android Studio

Unity Game Design

I HAVE VERY WELL VERESD WITH UNITY C# as i developed xr software like Vitro to accelrate xr education.

Backend Development

Labore velit culpa adipisci excepturi consequuntur itaque in nam molestias dolorem iste quod.

What I Did ?

Portfolio

AI Powered Autonomous Drone System

Inspired by E.D.I.T.H. from Spider-Man: Far From Home, I built my own AI-powered drone system using a fine-tuned multimodal model (LLaMA 3 7B + ViT) hosted on azure and accessed by the Raspberry Pi Zero. The drone captures live video, sends it for analysis, and can be controlled manually or autonomously via Flask endpoints using Ngrok for connectivity. It responds to user commands or real-time events—like detecting a flood, gathering online data, and navigating to the location using GPS. The ViT captions the video feed, which the LLM uses to decide the drone's actions and generate mission reports.

Fine Tuned LLM model

I have fine tuned tons of llm and vit model like llama 2 7b, Llama 3 7b, Granite etc for various task like code generation, task summerazation, text generation, market analysis, image captioning etc. All of those model are avaialable on on my hugging face repo.

My own Language Model Archtecture

I have constructed a new architecture without matrix multiplication that reduces computational cost and memory utilization. I had used ternery weight {-1,0,1} to replace the matrix operation on a llm archtecture. Access the model from Hugging Face.

Space AR - Solar System Tour

Space Ar is an augmented reality app that helps users explore the solar system. I had used Unity using C# to develop this app. The app allows users to view planets in 3D and learn about their characteristics. It also includes a feature that lets users take pictures of the planets and share them on social media.

SDRAM

SDRAM was my project for 2023's Nasa Space App chellenge where I developed an archtecture for people to communicate during disaster situation and also get disaster update. The archtecture uses lora 432 Mhz module creating a mesh topology with other lora module in the range of 12 km connecting to my android studio app (to access the native feature of android os) which provide a wide range of communication sector which increses as new node or lora module connect to the network. I have also trained a DNN model to predict lghtening which I hosted on azure computing storage and creating api to call from my android studio app. This archtecture made me the global finalist in that hackathon.

Electroinc Health Record

This Electronic Health Record (EHR) management system was developed using Android Studio in Java. It allows for the storage and management of patient data, including medical history, medications, allergies, lab results, and more. Prescriptions can be uploaded and converted to text using OCR, then analyzed using Gemini. Most notably, I trained a ViT (Vision Transformer) model for X-ray report analysis, which users can access directly from the app. The system is designed to be user-friendly and efficient, enabling healthcare providers to easily access and update patient records. It also includes features such as appointment scheduling, prescription management, and billing integration, making it a comprehensive solution for healthcare organizations.

Vitro- a vr learing platform in Unity(C#)

Vitro is like a special virtual reality (VR) world (platform) just for learning. It's a place where students and teachers can come together in VR to study and learn collaboratively. In the Vitro store, users can download tools like VR Chemistry Labs, 3D viewers, AI PDF viewers, Virtual Computers, AI voice notes, and more to help them with their studies.
Skills: Android Development · Application Development · Artificial Neural Networks · Computer Science · Convolutional Neural Networks (CNN) · Django · Flutter · Java · Machine Learning · Neural Networks · Python (Programming Language) · Research · Systems Design · Virtual Reality (VR)

Hear Helper- AI-powered assistive listening app

Hear Helper is an AI-powered assistive listening app for people with hearing impairments. Unlike typical transcription apps, it not only transcribes and summarizes conversations using Whisper and ChatGPT, but also detects and localizes critical environmental sounds—like alarms or crying babies—with directional and decibel data. This unique Sound Capture feature provides users with a deeper awareness of their surroundings, boosting both safety and confidence.

Ai Marketplace

I have been fine-tuning various LLM models like LLAMA 2, LLAMA 3, GRANITE, and GPT for tasks such as code generation and text summarization. This inspired me to create an AI marketplace same to an app store for AI models, where users can access my fine-tuned models and third-party models from Hugging Face. The platform features a playground tab with drag-and-drop functionality, allowing users to combine multiple models into custom pipelines for automating complex workflows. It also includes tools like text-to-speech, speech-to-text, and pdf-to-text. Users can export their custom pipelines as executable files (APKs or other formats) for local execution on devices like Android.

Hackathons and competitions where I participated

Accomplishment

August 25, 2023 - Jan 25, 2024

NASA Space Apps Challenge 2023

NASA Space Apps Challenge is world's largest hackathon where I developed SDRAM a hardware and software solution that allowed people to communicate during disaster when the netwrok infrstucture are broken. The solution uses lora 432 Mhz module to create a mesh topology with other lora module in the range of 12 km connecting to my android studio app (to access the native feature of android os) which provide a wide range of communication sector which increses as new node or lora module connect to the network. I have also trained a DNN model to predict lghtening which I hosted on azure computing storage and creating api to call from my android studio app. This archtecture made me the global finalist in that hackathon.

View Project

Jan 25, 2023 - Jan 25, 2024

NASA Space Apps Challenge 2024

Perticipated again on Nasa Space App Challenge again in 2024 and became a global nominee. I built an AR app in Unity (C#) for exoplanet exploration to make learning fun for kids. The app features 3D object interaction, solar system visualization, and exploration of celestial bodies.

View Project

Jan 25, 2023 - Jan 25, 2024

Pale Blue Dot Challenge

It was a visualization challenge organized by NASA, where my team and I visualized lightning-prone regions around the world using the Three.js library. We ranked 48th globally

View Project

Feb 10, 2023 - May 10, 2023

Microsoft Imagine Cup 2023

Microsoft Imagine Cup: In the Microsoft Imagine Cup 2023, I served as the team lead and lead technologist, securing $10,000 in funding for my startup, QbitLab. Our MVP product was Vitro, a VR learning platform (Unity/C#, AI, Azure) where students and teachers collaborate. Its store offers tools like VR Chemistry Labs, 3D viewers, AI PDF readers, virtual computers, and AI voice notes to enhance learning.

View Project

March 15, 2023 - April 20, 2023

IBM TechXchange Pre Conference watsonx Hackathon

I became top 100 global finalist @ IBM TechXchange Pre-Conference watsonx Hackathon.During the hackathon, I fine-tuned open-source LLMs like Llama 3, Granite, and GPT to boost productivity. I built a React Native app hosting these and third-party models, creating a Play Store-like marketplace where users can drag and drop models and tools to build pipelines.

View Project

May 5, 2024 - June 5, 2024

Microsoft Imagine Cup 2024

Perticipated again in 2024 and built an application Called Hear Helper which is an AI-powered assistive listening app for people with hearing impairments. Unlike typical transcription apps, it not only transcribes and summarizes conversations using Whisper and ChatGPT, but also detects and localizes critical environmental sounds—like alarms or crying babies—with directional and decibel data. This unique Sound Capture feature provides users with a deeper awareness of their surroundings, boosting both safety and confidence.

View Project

Jobs and Activities

Job and Activities

August 25, 2023 - Jan 25, 2024

QbitLab

QbitLab, a startup I co-founded and currently lead as its CEO. Founded in 2023, QbitLab is a tech startup focused on technology, information, and education. Under my leadership, my team delivered solutions via Fiverr and developed SaaS products in sectors such as healthcare, gaming, environmental sustainability, education, and XR, generating $20K in revenue. These products have an active user base of 15,000 customers.

Visit QbitLab

Jan 25, 2022 - Jan 25, 2024

Fiverr

I were employed in Fiverr as an cross platform app developer with react native, native app developer with android studio {Java} and an AI Engineer.

View Project

Jan 25, 2023 - Jan 25, 2024

MUGAS

I have served as a mentor and advisor for the Metropolitan University Geography and Astronomical Society (MUGAS) over the past year, where I had the privilege of mentoring university students in mathematics and machine learning with visualization.

MUGAS

Feb 10, 2023 - May 10, 2023

Future Talent Programme Bootcamp

I have been actively teaching as a mentor in the Future Talent Program (FTP) Bootcamp, where I have guided students from various educational institutions in my community. I taught the basic syntax of programming languages using C++, through hands-on DIY projects with microcontrollers like Arduino and NodeMCU, while also working to advance STEM education among underprivileged students.

VISIT FTP

Feb 10, 2023 - May 10, 2023

Aboard Station

I have been employed at Aboard Station for the past year, where I was primarily responsible for developing solutions for the online appointment system, managing social media ads, and overseeing the company’s website.

Visit Aboard Station

What Think Client About Me ?

Testmonial

Download free bootstrap 4 landing page, free boootstrap 4 templates, Download free bootstrap 4.1 landing page, free boootstrap 4.1.1 templates, meyawo Landing page

This was a wonderful experience working with Sakib. He built what was needed and took time to understand our requirements. I recommend him for your project.

morganwilliams2

Lorem ipsum dolor sit amet, consectetur adipisicing elit. Eaque doloribus autem aperiam earum nostrum omnis blanditiis corporis perspiciatis adipisci nihil.

Emily Reb

Blog/Research Paper

Reimagining LLMs with -1,0 & 1

By: Sakib 234 123

Any idea on what is the most costly operation on an LLM model? It’s matrix multiplication, which is the most dominant operation in most neural networks, including vector-matrix multiplication in the dense layer as well as matrix-matrix multiplication in the self-attention mechanism.

what if we could construct a new architecture without matrix multiplication that reduces computational cost and memory utilization while preserving the expressiveness of the network?

In this post, you will find out how I constructed a new architecture without matrix multiplication that reduces computational cost and memory utilization.

A Deep Neural Network Approach for HighlyEfficient Lightning Prediction and WarningSystems.

By Sakib 456 264

The absence of accessible and accurate early warning systems for lightning strikes poses a critical challenge, leading to severe consequences such as loss of life. Addressing this issue requires an innovative, integrated solution. My study investigates the development of a highly accurate lightning prediction model using predictive artificial intelligence, specifically Deep Neural Networks (DNNs), and essential meteorological parameters. This paper also has been accepted for presentation at the prestigious 8th International Conference on Engineering Research, Innovation and Education (ICERIE 2025).

About Me

Service

AI Engineer/ Data Sciencet

Cross Site App Developer

Unity Game Design

Backend Development

Full Stack Development

Cross Platform & Native App Developer

Unity Game Design

Backend Development

Portfolio

AI Powered Autonomous Drone System

Fine Tuned LLM model

My own Language Model Archtecture

Space AR - Solar System Tour

SDRAM

Electroinc Health Record

Vitro- a vr learing platform in Unity(C#)

Hear Helper- AI-powered assistive listening app

Ai Marketplace

Accomplishment

NASA Space Apps Challenge 2023

NASA Space Apps Challenge 2024

Pale Blue Dot Challenge

Microsoft Imagine Cup 2023

IBM TechXchange Pre Conference watsonx Hackathon

Microsoft Imagine Cup 2024

Job and Activities

QbitLab

Fiverr

MUGAS

Future Talent Programme Bootcamp

Aboard Station

Want to work with me?

Testmonial

morganwilliams2

Emily Reb

Blog/Research Paper

Reimagining LLMs with -1,0 & 1

A Deep Neural Network Approach for HighlyEfficient Lightning Prediction and WarningSystems.

Contact Me