Work, stack, what they are building. The agent answers from this card.
Raaghav Dhiman is a Computer Engineering student with a strong focus on Artificial Intelligence and Machine Learning. He has hands-on experience in building practical AI solutions, including fraud detection systems and retrieval-augmented generation pipelines. His technical expertise spans Python, Java, and various machine learning frameworks such as Scikit-learn and XGBoost. Currently, he is engaged in developing innovative projects like the Asset AI Copilot and a PDF Q&A chatbot, while continuously enhancing his skills through hackathons and competitions.
“Attended the AI Hackers Collective x Hacktoberfest Hack Day organized by Major League Hacking and AI Hackers Collective. Participated solo for the very first time and amidst all the nervousness and "I dont know what to do", figured out my way through the hackathon. I built the Asset AI copilot that basically works on two principles - 1. Managing and structuring data so that the models have proper context. 2. Building a generative UI so each user gets answers in the form that suits their questio…”
“Built a RAG (Retrieval-Augmented Generation) chatbot from scratch this week. Upload any PDF, ask questions, get answers grounded only in that document. The core problem RAG solves: a raw LLM either hallucinates or says "I don't know" about documents it was never trained on. RAG fixes this by retrieving the relevant chunks at question time and forcing the model to answer only from those. What surprised me most: RAG doesn't automatically stop hallucination. The model still "knows" things from tra…”
“Just wrapped up a fraud detection project ,end to end, from raw transaction data to a live deployed app. The dataset: 284,807 credit card transactions, only 0.17% of them fraudulent. That single number shapes every decision in a project like this. A model that predicts "not fraud" for everything gets 99.8% accuracy while catching zero fraud. So accuracy is thrown out early, and precision/recall become the real language of the problem. The biggest lesson wasn't a model or a metric ,it was realiz…”
“Built an IPL Playoff Predictor that predicts match winners based on how teams actually perform. This is a Rule-Based AI Engine that predicts the winner based on multiple factors that defines a team's performance, like - Batting Score - Runs scored in different phases of the match like powerplay, death, middle. Then strike rate at the death overs , dependency on top 4 batters. Bowling Score - Wickets in the powerplay, death, middle overs. Economy during death overs , dot ball percentage. Form -…”
“From acting as Akshay Kumar during our project pitch to qualifying for the second round, the MIT Kurukshetra Hackathon was so great. 36 hours of pure innovation, zero sleep, and making the brain work overtime made it one of the most fun and intense experiences of my life. Also, had an incredible team that was as amazing in work as they were in fun. #Hackathon #Innovation #MITKurukshetra2025 #Teamwork #LearningByDoing”
“Here goes nothing—first post! Took part in Code Clash (SCOE Pune), Techtonic Technomela Hackathon, and CipherX AI Cybersecurity Challenge—first steps count as wins. Steady progress. Feb–Apr 2025. #FirstPost #StudentDeveloper #Coding #Hackathon #Cybersecurity #LearningInPublic #Pune”
AI
Machine LearningAIMachine Learning
LangChain
StreamlitMachine LearningStreamlit
Data AnalyticsStreamlit
Machine LearningAdvanced
SQLMid
HTML/CSSMid
Data Structures & AlgorithmsMid
ML / AIMid
Software DevelopmentMid