🎓 Pre-Final Year Student @ IIT Roorkee | 🤖 AI Enthusiast | ✍️ Tech Blogger
- 🎓 Pre-final year undergrad in Industrial Engineering (Mechanical & Industrial Engineering Dept.) @ IIT Roorkee.
- 🤖 Passionate about Machine Learning, Deep Learning, NLP, CV, and Large Language Models (LLMs).
- 🧩 Research interests: AI red teaming(Jailbreaking LLMs, Text-to-Image jailbreak attacks), RLHF(PPO,GRPO), agentic AI, statistical learning, and LLM Alignment.
- ✍️ Love sharing knowledge via blogs to help others on their ML journey.
- 🏊 Outside academics: swimming, puzzles, and engaging discussions keep me active.
- Presently researching on Jailbreaking of Audio Language models and adversarial attacks on Autoregressive Text-to-image models and robust Watermarking techniques/algorithm.
- 📌 Present (2025): Exploring internships in ML & Data Science for Fall 2025 / Summer 2026.
- ✅ Mar 2025: Completed Research Internship at UiT The Arctic University of Norway.
- 📄 Research on Time Series Forecasting & Anomaly Detection of Rolling Bearings accepted at ASME, Canada.
- 🎯 Jan 2025: Selected as Research Intern @ UiT The Arctic University of Norway.
- Languages: Python, C++, MySQL
- Frameworks: PyTorch, TensorFlow
- LLMs & NLP: HuggingFace, LangChain, CrewAI, PEFT (LoRA/QLoRA), RLHF (PPO, GRPO, DPO)
- Other Tools: Git, Docker, Jupyter, FastAPI
🔗 Explore more on my GitHub Repos
- 🎨 Drawing with LLMs → Improving SVG code generation using QLoRA + RLHF (GRPO).
- 🌾 Agri-Agent → Multi-agent framework (CrewAI) for pest/weed detection with retrieval + web search.
- 📑 Multimodal QnA Chatbot → Chat with research papers using text + image summarization (Llama 3 + Gemini 2.5).
- 🚁 Drone Fault Classification → Multi-task CNN on drone audio achieving 95% accuracy.
- 🌐 Neural Machine Translation → Seq2Seq Transformer for English–Italian with BLEU 36.
- 🔬 Research Intern @ UiT The Arctic University of Norway (2025)
- Built anomaly detection model for rolling bearings using Attention-LSTM + Physics-informed loss.
- Research presented at ASME QNDE 2025, Montreal.
- 💻 Machine Learning Intern @ STYAVA (2024)
- Fine-tuned Whisper for Indic ASR, achieving 30% WER on Malayalam/Hindi.
- Built a data framework for 12K+ audio samples and deployed testing with FastAPI.
- 🥇 Ranked 234/4000+ in Kaggle Competition Drawing with LLMs.
- 🥈 Top-200 in Shell AI Hackathon among 10,000+ participants.
- 📑 Research work presented at ASME 2025 (Canada).
- 🎯 Solved 300+ problems on LeetCode (Handle:
kolbe_electrolysis).
