|
I'm Mahir Faysal Tusher, a Computer Science and Engineering student at Chandpur Science and Technology University (CSTU), Bangladesh. My career goal is to work in the AI ecosystem, with a particular focus on Machine Learning and AI Engineering. I'm actively building projects around LLM applications, AI agents, Python tooling, and data-driven software — with an emphasis on evaluation, reproducibility, and shipping things people can actually use. My research interests center on AI for scientific discovery: AI applications in agriculture, data-driven molecular and drug discovery, and energy-based systems. I care about the open-source community and am working toward contributing to it consistently. |
|
🗺️ How it all connects — from foundations to discovery
flowchart LR
subgraph Foundations["🎓 Foundations"]
direction TB
CS["CS + Math<br/>CSTU coursework"]:::foundation
DSA["Algorithms + Systems"]:::foundation
end
subgraph Core["🧠 Machine Learning"]
direction TB
ML["Modeling + Evaluation"]:::research
DL["Deep Learning + LLMs"]:::research
end
subgraph Build["⚙️ AI Engineering"]
direction TB
Apps["LLM Applications"]:::applied
Agents["AI Agents"]:::applied
Tools["Python Tooling +<br/>Data-Driven Software"]:::applied
end
subgraph Science["🔬 AI for Scientific Discovery"]
direction TB
Agri["🌿 Agriculture"]:::domain
Drug["🧬 Molecular + Drug Discovery"]:::domain
Energy["⚡ Energy-Based Systems"]:::domain
end
subgraph OSS["🌍 Open Source"]
direction TB
Contrib["Contributions +<br/>Collaboration"]:::community
end
CS --> ML
DSA --> ML
ML --> DL
ML --> Tools
DL ==> Apps
DL ==> Agents
Apps --> Agri
Apps --> Energy
Agents --> Drug
Tools --> Agri
Agri --> Contrib
Drug --> Contrib
Energy --> Contrib
classDef foundation fill:#102A4A,stroke:#36BCF7,stroke-width:2px,color:#EAF6FF;
classDef research fill:#1E1B4B,stroke:#8B5CF6,stroke-width:2px,color:#F5F3FF;
classDef applied fill:#052E2B,stroke:#22C55E,stroke-width:2px,color:#ECFDF5;
classDef domain fill:#3B2506,stroke:#F59E0B,stroke-width:2px,color:#FFFBEB;
classDef community fill:#0F2742,stroke:#60A5FA,stroke-width:2px,color:#EFF6FF;
style Foundations fill:#0F172A,stroke:#233044,stroke-width:1px,color:#CBD5E1
style Core fill:#111827,stroke:#312E81,stroke-width:1px,color:#DDD6FE
style Build fill:#0B1F1A,stroke:#14532D,stroke-width:1px,color:#BBF7D0
style Science fill:#1C1407,stroke:#78350F,stroke-width:1px,color:#FDE68A
style OSS fill:#111827,stroke:#1D4ED8,stroke-width:1px,color:#BFDBFE
linkStyle 0,1 stroke:#36BCF7,stroke-width:2.5px;
linkStyle 2,3 stroke:#8B5CF6,stroke-width:2.5px;
linkStyle 4,5 stroke:#22C55E,stroke-width:3px;
linkStyle 6,7,8,9 stroke:#F59E0B,stroke-width:2.5px;
linkStyle 10,11,12 stroke:#60A5FA,stroke-width:2.5px;
|
Machine learning and computer vision for crop health — turning field images and agronomic data into decisions farmers can act on. Applied in Cropora: plant-leaf disease classification on Android with cloud inference. |
Data-driven discovery — representation learning and property prediction that help prioritize which molecules are worth testing. Direction: reproducible pipelines that connect chemistry data, models, and rigorous evaluation. |
Simulation, forecasting, and LLM-grounded decision support for energy systems — where every number is computed first and only then explained. Applied in GridLens: microgrid scenario modeling with a RAG explanation layer. |
Tip
Working on AI for agriculture, molecular/drug discovery, or energy systems? I'd love to contribute — open an issue on one of my repositories or reach me through the links in Connect.
|
LLM-assisted energy intelligence platform: a deterministic microgrid simulation engine paired with a RAG explanation layer for scenario modeling, forecasting, and evidence-grounded decision support.
|
Android app for plant-leaf disease detection: leaf photos are analyzed through a FastAPI + TensorFlow cloud backend, with on-device TensorFlow Lite inference in progress.
|
|
Multi-backend AI chatbot with a customizable Gradio interface, supporting 100+ models across providers through one unified conversational workflow.
|
Machine-learning-powered web app that turns a trained model into an interactive heart-disease risk prediction interface.
|
|
End-to-end ML classification pipeline on clinical data — preprocessing, modeling, evaluation, and analysis in reproducible notebooks.
|
Versatile, multi-interface Python to-do application that adapts to your workflow — one task model, several ways to use it.
|
📚 More learning lanes and proof repositories
| 🏛️ Degree / Program | 🏫 Institution | 📍 Location | 📌 Status |
|---|---|---|---|
| B.Sc. in Computer Science and Engineering | Chandpur Science and Technology University (CSTU) | Chandpur, Bangladesh | In progress |
| Higher Secondary Certificate | Chandpur Govt College | Chandpur, Bangladesh | 2022 |
| Secondary School Certificate | Hasan Ali Govt High School | Chandpur, Bangladesh | 2020 |
📖 Structured learning tracks I document publicly
| 🧭 Track | 📁 Notes & exercises |
|---|---|
| Machine learning & deep learning | ML Specialization (Andrew Ng) · Hands-On ML · PyTorch for Deep Learning |
| LLM engineering & agents | LLM Engineering · Build an LLM From Scratch · AI Builder with n8n |
| Career tracks | Machine Learning Engineer · Associate AI Engineer for Developers |
| Python & data | CS50P · 100 Days of Code · Data Science in Python · Pandas Exercises |
| Research & study skills | How to Write a Successful Research Paper · Learning to Learn |
|
|
|
Open to: ML / AI engineering internships and early-career roles · research collaboration in AI for science · open-source contribution · LLM and agent projects · Python tooling · learning communities and mentorship conversations.
