SIH Round 2 Qualifier
Smart India Hackathon 2024
Qualified for the second round of India's largest hackathon, competing among thousands of teams nationwide.
AIML & Physics Enthusiast | Developer
Building innovative solutions at the intersection of Artificial Intelligence, Embedded Systems, and Physics.
const Mihir = {
name: "Mihir Mohite",
focus: ["AIML", "Physics", "Development"],
skills: ["Python", "MATLAB", "C++", "ML/DL", "IoT"],
goal: "Innovative Problem Solving"
};
A results-driven B.Tech student specializing in AI & ML with extensive hands-on experience in Large Language Models, Retrieval-Augmented Generation (RAG), and agentic AI systems.
Currently serving as President of CoDeC and Machine Learning Intern at BeyondBot Technology, I have a proven track record of architecting production-grade ML solutions including enterprise knowledge graphs, multi-modal document processing pipelines, and scalable cloud-native applications.
My expertise spans the full ML lifecycle from research to deployment, with a strong foundation in modern AI frameworks (Vertex AI, Gemini API, Agno), cloud infrastructure (AWS, Google Cloud), and full-stack development. I specialize in building RAG systems, implementing Computer Vision and OCR pipelines, and developing knowledge graph solutions with Neo4j.
What drives me: I'm passionate about applying theoretical knowledge to solve real-world challenges at the intersection of AI, embedded systems, and physics. Whether it's implementing SLAM algorithms for autonomous rovers, forecasting energy loads with deep learning, or building enterprise AI platforms, I thrive on creating innovative solutions that make an impact.
Key Highlights:
When not coding or studying, I enjoy exploring physics concepts, contributing to open source, and staying updated with the latest in AI research.
Electronics & Communication Engineering, Specialization in AI & ML
CGPA: 8.5/10.0 | Currently in 3rd Year | Expected Graduation: May 2027
Developing and optimizing production-grade AI/ML solutions with focus on LLMs and enterprise applications.
Contributed to production software systems with focus on ML-based validation and data quality.
Qualified for the second round of India's largest hackathon, competing among thousands of teams nationwide.
Secured second place in the hardware and circuit design competition, demonstrating expertise in embedded systems.
Advanced to the final rounds in this prestigious AI and data science competition, demonstrating strong analytical and machine learning skills.
Qualified for the final rounds of MIT-WPU's flagship hackathon, developing innovative AI/ML solutions under time constraints.
Leading the strategic direction and technical initiatives for university's premier programming club.
Spearheading AI/ML project initiatives from conception to completion.
Engineered enterprise knowledge graph generation system supporting structured (SQL), semi-structured (JSON), and unstructured (TXT) data sources with Neo4j and Gemini API for intelligent entity extraction.
Built production-grade RAG system using Sentence Transformers for tokenization, chunking, and embedding with Pinecone vector database. Implemented alternative pipeline using IBM Docling with multiple chunking strategies.
Trained CNN on Common Crawl dataset using FFDNet (Form Field Detection Network) for automated form field detection and extraction from diverse document types.
Developed a rover capable of navigating hazardous areas using LiDAR-based SLAM for mapping (MATLAB) and equipped with Temperature, Humidity, Ultrasonic, and MQ5 gas sensors for environmental monitoring.
Implemented an MLP (Multi-Layer Perceptron) deep learning model to forecast electricity demand in Delhi, aiding in resource management.
Designed and built a solar tracking system using Light Dependent Resistors (LDRs) to automatically orient solar panels towards the sun for maximum energy capture.
Developed an emergency SOS tracker utilizing LoRa (433 MHz FM) communication for long-range, low-power distress signaling in areas with poor connectivity.
Created a system using LiDAR and MATLAB's Fast Fourier Transform (FFT) to non-contactually measure the rotational speed of fans.
Developed university-wide feedback platform with WhatsApp integration using Meta Business API. Features agentic AI workflow using Agno framework for dynamic question generation and adaptive surveying. Acknowledged by university administration.
Interested in learning more about my background and experience?
Download my complete resume to get a comprehensive view of my skills, projects, education, and achievements.
AIML & Physics Enthusiast | Developer
Interested in collaborating or discussing tech? Let's connect!
I'm always open to new ideas, projects, and learning opportunities. Feel free to reach out via email or connect on social media.