I'm Rohit Kumar — a final-year CSE (AI/ML) student building production-grade LLM systems, RAG pipelines, and agentic orchestration layers. Currently at C-DAC Mohali under MeitY, Govt. of India.
Builder. Learner. Obsessive about shipping systems that actually work in production — not just notebooks.
I write code the way engineers write infrastructure — with intention. My work lives at the intersection of large language models, retrieval systems, and applied ML infrastructure.
At C-DAC Mohali under MeitY (Govt. of India), I'm designing secure RAG pipelines with guard-layer mechanisms against prompt injection and data leakage — the kind of work where correctness isn't optional.
Beyond the stack, I'm a finalist at brAInwave (DTU) and a state-level participant at the OpenAI Academy × NxtWave Buildathon. Currently sharpening fundamentals in transformer internals, fine-tuning techniques, and MLOps — and always open to deep, hard problems.
Some I've mastered. Some I'm actively learning. All honest.
Real production work. Real impact. Real stakes.
Production systems, not class assignments.
A decentralized orchestration layer for managing AI agents at scale. Built an LLM-based planner that generates execution plans and routes user queries to specific agents via signature validation. Vector-based agent discovery powered by Weaviate, with verifiable metadata storage.
View on GitHub →GenAI-powered OCR pipeline for automated document verification. Extracts structured data from unstructured documents and produces cryptographic hashes for integrity — reducing manual verification time by 60%.
Real-time fraud detection system using ML classification on transaction streams. Feature-engineered pipelines, ensemble classifiers, and low-latency inference — a brAInwave DTU Top-10 finalist project.
End-to-end ML system for predicting logistics delays using time-series features and geotagged shipment events. Reduced delivery-fraud discrepancies significantly across real logistics flows.
Move your cursor over the canvas to activate neurons. A live interactive demo of how signal propagates through a simple feed-forward network.
Consistent practice. Public code. Every day.
Daily practice across arrays, graphs, dynamic programming, and system design. Focus areas: ML-adjacent problems, graph algorithms, and optimization.
View profile →DTU national hackathon — AI-powered fraud detection system selected among top 10 teams.
State-level participant in the Buildathon — rapid prototype of applied GenAI systems.
IIT Madras Machine Learning course — rigorous foundations in supervised and unsupervised methods.
Certification covering transformers, prompt engineering, and RAG fundamentals.
Selected for Government of India research lab — working on secure, production-grade RAG.
16+ GitHub repositories. Every project shipped, documented, and production-ready.
I'm actively looking for full-time AI Engineer roles starting mid-2026. If you're working on LLM infrastructure, agentic systems, or applied ML — I'd love to talk.
rohitpkkumar10@gmail.com ↗