Embedded Files
Undergraduate Projects
Undergraduate Projects
Supervisor: Dr. Mohammad Arif Hasan Mamun
Designed and fabricated a heat exchanger with novel baffle designs to enhance thermal efficiency.
Conducted CFD and thermal simulations (ANSYS, SolidWorks, Python); validated results with hydraulic testing.
Abstract: Gasketed Plate Heat Exchangers (GPHX) are a relatively newer version of heat exchangers that gained popularity due to having some distinct advantages over its other traditional counterparts, e.g. Shell and Tube Heat Exchangers (STHX) and Double Pipe Heat Exchangers (DPHX). The unique advantages of GPHX include: Compact design requiring less space, higher area to volume ratio, high heat transfer efficiency, ease of modifying the performance by simple addition or removal of plates, quick and easy detection of leakage. Industries that require stringent process control and are extremely sensitive to leakage, like pharmaceutical industries, food processing plants, chemical industries etc. can therefore easily make GPHX their ideal choice of heat exchanger. Typically, the modern GPHXs are made with corrugated plates, which help in inducing turbulence in the flowing fluid, thus enhancing the heat transfer. However, the cost of corrugation of the plates can be a challenge. Especially in smaller scale projects which do not require a batch production, creating small corrugation features can be a huge economic barrier. Hence, as an alternative to this, the idea of creating GPHXs with rubber baffles over flat plates came up. The baffles essentially do the same thing corrugation does, enhance the heat transfer by agitating the flowing fluid, while also taking significantly lower production cost. In this project, we investigated the performance of GPHXs with rubber baffles made out of the same material as the gasket (silicon rubber) and made a comparative analysis between two designs of the baffles (vertical and horizontal). We also manufactured a GPHX with vertical baffles and identified the challenges, advantages and future scopes of such a heat exchanger. The results of this study could provide insights to the feasibility of using rubber baffles in GPHXs over corrugation for smaller scale projects, and add to the ongoing research to optimize the performance of heat exchangers for various industrial applications.
Supervisor: Dr. Kazi Arafat Rahman
Designed and modeled 3-DOF legs in SolidWorks and fabricated parts via 3D printing; implemented a trot-gait algorithm on Arduino for quadruped locomotion.
Integrated ESP32-CAM and Bluetooth modules for live video streaming and wireless control.
Personal Projects
Personal Projects
774-Million-Parameter Large Language Model from Scratch Jan 2025
Designed and trained a custom GPT architecture (multi-head attention, causal masking, GELU-activated feedforward layers, layer normalization) in PyTorch, with BPE tokenization and token/positional embeddings.
Fine-tuned the model for spam classification and performed instruction fine-tuning (1,100 Alpaca-format pairs); evaluated against LLaMA 3.1 (8B) achieving 62.7% accuracy and deployed via the Ollama REST API.
GitHub: naimulnashid/LLM_from_Scratch
Training of an LLM Model build from Scratch
Instruction Fine-tuning of Pre-trained Weights from OpenAI
Classification Fine-tuning of Pre-trained Weights from OpenAI
Instruction Fine-tuning of Pre-trained Weights from OpenAI
Context-Aware Reasoning App Aug 2024
CSV Query: Provides coherent answers to frequently asked questions using CSV files as a knowledge base.
SQL Query: Translates natural language questions into MySQL queries for database execution.
PDF Query: Conducts question-and-answer tasks based on information extracted from provided PDFs.
ATS: Matches resumes with job descriptions, highlighting missing key points to improve resumes.
Built a context-aware reasoning application using LangChain with Gemini-Pro API, integrating FAISS/Chroma vector databases for information retrieval.
Applied few-shot learning, retrievers, and structured output parsers to enhance LLM query handling and deployed the application via Streamlit.
GitHub: naimulnashid/LangChain_Projects
Real-Time Gesture-Based Volume Control Apr 2024
Designed a computer vision application with OpenCV for controlling system volume through hand-gesture recognition.
Applied MediaPipe’s landmark detection and distance mapping for smooth real-time control.
First-Person Shooter Game Apr 2024
Basic "Flappy Bird" Game Apr 2024
Flamethrower Fire Apr 2024
Open Source Projects
Open Source Projects
AI Usage Tracker Aug 2026
Data Usage Tracker Aug 2026
Screen Time Tracker Aug 2026
Internet Speed Meter Sep 2026
Music Vault - Cross Platform Oct 2026
Page updated
Google Sites
Report abuse