Final-semester Computer Science undergraduate at BRAC University
Backend Engineering • Applied AI • Full-Stack Development • Open Source

I am a final-semester Computer Science undergraduate at BRAC University, Bangladesh, expected to graduate in September 2026.

My primary interests are backend engineering, full-stack development, applied AI, developer tools, and data-driven systems. I learn by building practical software and working across the development lifecycle, including API design, database modeling, testing, documentation, containerization, and continuous integration.

My recent work includes a grounded RAG system for codebase understanding, a full-stack clinic operations platform, a MERN student marketplace, a database-backed alumni networking platform, and an undergraduate machine learning thesis on apartment price prediction in Dhaka.

I also contribute to open-source projects and have merged contributions to mem0 and Agno.

• Building backend and applied AI systems
• Working with RAG, LLMs, vector search, BM25, and reranking
• Developing APIs with FastAPI, Django REST Framework, and Node.js
• Building full-stack applications with React and TypeScript
• Improving software quality through testing, Docker, CI/CD, and clean engineering practices
• Contributing to open-source projects
• Strengthening data structures, algorithms, and problem-solving skills

Languages: Python, SQL, JavaScript, TypeScript, C, C++
Backend & Frontend: FastAPI, Django REST Framework, Node.js, Express.js, React, Tailwind CSS, Vite
Databases & AI: PostgreSQL, MongoDB, MySQL, Milvus Lite, RAG, LLMs, BM25, Vector Search, Reranking, SHAP
Tools: Git, Docker, GitHub Actions, CI/CD, pytest, REST APIs, Linux

Added Google Cloud Vertex AI embedding provider support to the mem0 TypeScript SDK through a merged upstream pull request.

The contribution covered provider implementation, configuration support, unit tests, and documentation.

Tech: TypeScript, Google Cloud Vertex AI, Embeddings, SDK Development
Pull Request: mem0ai/mem0#5882

Contributed a merged pull request that improved CustomApiTools authorization-header handling to prevent unintended API key forwarding when no base URL is configured.

The fix was backed by unit tests covering the affected authorization behavior.

Tech: Python, pytest, API Security
Pull Request: agno-agi/agno#8582

A local-first RAG system that indexes GitHub code, documentation, and issue/PR history to answer codebase questions with grounded citations to files, line ranges, and issues.

Tech: Python, FastAPI, Milvus Lite, Ollama, React, TypeScript, BM25, Reranking
Highlights:

• Dense vector search and BM25 retrieval
• Reciprocal Rank Fusion
• Cross-encoder reranking
• Grounded answers with source citations
• 146 unit tests
• Strict mypy and ruff checks
• Docker and GitHub Actions CI

Repository: CodeCompass

A full-stack clinic operations platform with role-based workflows for Admin, Front Desk, Triage Nurse, Doctor, and Diagnostics users.

Tech: Django REST Framework, PostgreSQL, React, TypeScript, Docker, Tailwind CSS
Highlights:

• Role-based operational workflows
• Priority queueing
• Wait-time estimation
• Diagnostic flow planning
• Staff overrides
• Audit logging
• Automated tests
• Docker Compose setup
• Synthetic demonstration data

Repository: CareFlow AI

A MERN marketplace developed by a four-member team for university students. I led the team and implemented the buyer-seller communication functionality.

Tech: MongoDB, Express.js, React, Node.js, Redux Toolkit, JWT, Multer
Highlights:

• JWT authentication through HTTP-only cookies
• bcrypt password hashing
• Role-based student and admin access
• Buyer-seller messaging
• Unread-message tracking
• Reporting and moderation workflows
• Seller-scoped product CRUD
• Search, filtering, and pagination
• Multer image uploads

Repository: Campus Cart

A secure alumni networking platform built on a normalized MySQL schema with role-specific functionality.

Tech: PHP, MySQL, HTML, CSS
Highlights:

• Role-specific search
• Messaging
• Event and mentorship workflows
• Prepared statements
• Password hashing
• CSRF protection

Repository: BRACU Alumni Connect

Undergraduate thesis at BRAC University focused on predicting apartment prices in Dhaka using structural and location-based features.

Built a stacking ensemble on 9,167 cleaned apartment listings, achieving:

• R² = 0.867
• 14.59% MAPE on a held-out test set
• SHAP-based model interpretability
• Split-conformal prediction intervals

Tech: Python, scikit-learn, XGBoost, LightGBM, SHAP

• Solved 700+ problems across LeetCode and Codeforces
• Focused on data structures, algorithms, and competitive programming