Building intelligent systems at the intersection of Generative AI, RAG pipelines, and agentic workflows. Currently engineering enterprise-grade AI solutions at Montra Electric (Murugappa Group).
I'm V.S. Karthik Manikandan, a Full Stack LLM Engineer from Chennai with a relentless passion for building AI systems that solve real-world problems. My journey started at Vellore Institute of Technology (VIT), where I graduated with a B.Tech in Computer Science in 2025 — and I haven't stopped shipping since.
Currently at Montra Electric (Murugappa Group), I architect the AIRA platform — an enterprise-grade GenAI chatbot powered by RAG, LLM, MCP, LangGraph, LangChain, and LangSmith. I integrate 6+ data sources, optimize vector search pipelines, and deploy agentic AI workflows that have boosted data accessibility by 70% and cut retrieval time by 95%.
What makes me different? I don't just build models — I build products. From a Voice-Enabled Sign Language Recognition system achieving 95% accuracy to an iOS Watch App built with Cursor AI (a first-of-its-kind AI+Human collaboration in India at the time), I push boundaries where AI meets human impact.
I'm also a content creator who completed 200 consecutive daily coding challenge videos on YouTube (2025–2026), covering Data Structures, Algorithms, ML, and AI — proving that consistency compounds. My research paper on Neural Summarization for Low-Resource Indic Languages is currently awaiting publication.
Beyond code, I've held an Elite World Record (2019), coordinated sponsor teams for Riviera 2025 (VIT's world-renowned cultural fest), and sharpened my public speaking at Toastmasters (2024–2025). I believe the best engineers are also great communicators.
Achieved in 2019 during school years
Neural Summarization for Indic Languages — awaiting publication
Daily coding challenges, 2025–2026
Public speaking & event coordination at VIT
From enterprise AI platforms to banking systems — building production-grade software.
From research papers to production AI systems — every project tells a story of solving real problems.
Hybrid Neural Summarization built on top of MuRIL Enhanced Graph Ranking and LoRA Optimized IndicBART. Tackles the critical challenge of text summarization for under-resourced Indian languages using state-of-the-art NLP techniques.
Full-stack GenAI platform at Montra Electric integrating 6+ enterprise data sources. Built with RAG, LLM, MCP, LangGraph, LangChain, and LangSmith. Boosted data accessibility by 70% and reduced retrieval time by 95%.
System and Method for Autonomous Risk-Based Evaluation, Retraining, and Deployment Control of LLMs. A pioneering innovation project addressing critical AI safety challenges in production LLM systems.
AI-powered music app feature using Model Context Protocol (MCP). Built with 80% Cursor AI and 20% human interaction — demonstrating the power of AI+Human collaborative development.
Full-stack application built with Google Gemini and LangGraph, demonstrating end-to-end GenAI application architecture with modern frameworks.
Full-stack RAG pipeline using LangChain and Llama 3.2 for real-time fake news detection. Integrated FAISS vector database with HuggingFace embeddings, achieving 20% accuracy improvement and 15% reduced query latency.
Image classification system for road pothole detection using CNN, ResNet & EfficientNet models. Comparative analysis of deep learning architectures for real-world infrastructure monitoring.
Interactive AI chatbot built with Meta's Llama 3 and Streamlit framework. Clean UI for conversational AI with streaming responses and context management.
Innovative Apple Watch app to browse and watch reels on iWatch. Built using Cursor AI — one of the first AI+Human collaborative iOS projects in India, pioneering a new development paradigm.
Achieved 95% gesture detection accuracy using CNN-based deep learning model with voice output via gTTS. Enhances accessibility for hearing-impaired users with a 30% optimized recognition pipeline.
Deep learning model for classifying multiple skin disease categories from dermatological images using Convolutional Neural Networks. Aimed at assisting early diagnosis and dermatological screening.
Machine learning system using Random Forest Trees for early detection of autism in children. Focused on feature engineering and model interpretability for sensitive healthcare applications.
Comprehensive collection of Machine Learning projects covering supervised and unsupervised learning, deep learning, and GenAI. Proficient in Exploratory Data Analysis (EDA), classification, regression, clustering, and neural network architectures.
The tools and technologies I use to build intelligent systems.
Continuous learning is my unfair advantage.
Vellore Institute of Technology (VIT), Vellore
Technobytes — IITM Research Park
Agnirva
SLA Institute — Comprehensive EDA, Supervised & Unsupervised Learning, Neural Networks
Achieved an Elite World Record during school years — a testament to early ambition and excellence.
2019Coordinated sponsor teams for VIT's Riviera, one of the world's top cultural festivals, managing partnerships and logistics.
2025Active member honing public speaking, leadership, and communication skills through regular speaking engagements.
2024 – 2025Participated in tech talks and discussions at VIT's technology-focused club, sharing knowledge on emerging tech trends.
2022Interested in collaborating on AI projects, discussing research, or just saying hi? I'd love to connect.
I'm always open to new opportunities, collaborations, and conversations about AI, technology, and innovation.