I am a Pentester with a passion for ethical hacking, machine learning, LLM Agents and solving complex problems through code. I love exploring new technologies and applying my skills to real-world projects.
- Languages: Python, Java, C
- Libraries & Frameworks: Scikit-Learn, Pandas, NumPy
- Tools: Git, Jupyter Notebook, VS Code
- Areas of Interest: Ethical hacking, Machine Learning, Natural Language Processing
- Description: An autonomous LLM agent that plays the browser strategy game Travian Legends end-to-end. The agent teaches itself the game from the official guides, builds its own automation on the fly, maintains a persistent memory across stateless sessions, and develops genuine long-term strategy — all from a single high-level goal. An exploration of long-horizon, self-improving agentic systems in a real (sandboxed) web environment.
- Tech Stack: Claude Code, Playwright, JavaScript, Python, Caddy, Markdown
- Repository: Travian Agent
- Highlights:
- Designed a continuous sense–decide–act–sleep loop driven by a single
AGENT.mdprompt, running unattended via Claude Code's/goal. - Built a self-growing library of injectable JavaScript snippets (sensors → atomic actions → routines → strategies) that the agent writes and extends as it plays.
- Engineered a fully file-based memory system (game state, strategic TODO, error log, evolving playbook) to give the agent long-term continuity across context-less sessions.
- Implemented adaptive sleep scheduling from in-game completion timestamps to cut token usage and produce realistic activity patterns.
- Compared model behavior (GPT 5.5 vs. Claude Fable 5), surfacing emergent long-term strategy
- Designed a continuous sense–decide–act–sleep loop driven by a single
- Description: A machine learning project aimed at identifying the authorship of WhatsApp messages. This project leverages natural language processing techniques and machine learning algorithms to analyze the writing styles of different authors and accurately attribute messages to their respective authors.
- Tech Stack: Python, Scikit-Learn, Pandas, NLTK, Matplotlib
- Repository: WhosApp
- Highlights:
- Implemented various text preprocessing techniques such as tokenization, stemming, and stop-word removal.
- Explored multiple machine learning algorithms including Naive Bayes, SVM, and Random Forest.
- Achieved high accuracy in authorship prediction through model evaluation and hyperparameter tuning.
- Visualized data distributions and model performance using Matplotlib.
- LinkedIn: My LinkedIn