Experience
AI/ML Trainee
Technosavvys Education Technology
- ▹Engineered a BERT-based text classifier via Hugging Face to automate sentiment analysis of 500+ student course reviews, achieving 88% accuracy through iterative hyperparameter tuning.
- ▹Streamlined instructional workflows by reducing manual review time by 60%, while simultaneously boosting baseline model accuracy by 15% through advanced NLP techniques.
- ▹Deployed a real-time Streamlit dashboard supported by a scheduled SQLite backend, providing 10+ mentors with live sentiment tracking across 5+ active courses.
Education
BCA with IBM Specialization (AI & Data Science)
United University Prayagraj
Certifications
Building LLM Applications with Prompt Engineering
NVIDIA Deep Learning Institute
2025- ▹Built LLM-powered applications using NVIDIA NIM (Llama 3.1) and LangChain with live API integration and Pydantic-structured output validation across a capstone agentic workflow.
Featured Projects
OMNICREW AI
An autonomous multi-agent B2B lead generation platform that scrapes, structures, and enriches web data in real time. Engineered with a resilient 3-tier LLM fallback router (DeepSeek → Mistral → Gemini) for 99.9% uptime, alongside autonomous Snov.io email enrichment and strict JWT multi-tenant data isolation.
Agents Online
Tasks Automated: 1,204
LumanGuide
An enterprise Agentic RAG system built on a LangGraph state machine, designed to streamline engineering onboarding and knowledge management. Orchestrates dynamic document grading and query routing with strict AppSec measures (JWT RBAC, prompt injection guardrails). Powered by DeepSeek for rapid chat generation and MongoDB Atlas for persistent, isolated multi-tenant chat history.
SaaS System Nominal
Socratic Engine Active
GitHub Activity
Open-source work and contributions
Contact Form
Please contact me directly at nileshraj1735@gmail.com or drop your info here.