Nikhilesh (Nik) Chhetri
AI Engineer · Tupelo, MS · chhetrinikhilesh@gmail.com · nikchhetri.com
A curious and motivated engineer who strives to provide value and learn.
Professional Experience
AI Engineer, Central Service Association
- Architected, developed, and deployed efficient LLM-based systems using vLLM and Docker, delivering secure and scalable AI applications that are currently offered to clients.
- Fine-tuned large language models using LoRA for cost-effective domain adaptation; integrated semantic search with vector databases to build high-accuracy, context-aware RAG pipelines.
- Migrated CI/CD pipelines from GitHub Actions to Jenkins, improving deployment reliability and increasing team productivity by 50%.
- Automated infrastructure provisioning and deployment using Ansible, streamlining production workflows and improving stability.
- Led technical decision-making and took ownership of architecture design and system deployments across AI, full stack, and infrastructure domains.
Student Researcher, Mississippi State University
- Researched travel time estimation for the City of Starkville using Torch, YOLO, and image processing on live traffic camera data, with LSTM and GNN models for travel time prediction.
- Researched the Highway-In-The-Sky (HITS) protocol for unmanned drones for last-mile delivery for Amazon and the Federal Aviation Administration (FAA) for the Amazon Prime Drone Delivery platform.
- Wrote documentation for the Mississippi Department of Transportation (MDOT) ATSPM implementation.
Education
Startup Foundry, Harvard Business School
Master's in Data Science, University of the Cumberlands
Bachelor's in Computer Science, Mississippi State University
Skills
AI & Machine Learning: LLMs, RAG, LoRA, AI Agents, AI Evaluation Strategy, Vector Databases, Torch
Software Engineering: Python, APIs, SQL and NoSQL, Web Crawlers, Full Stack Engineering, End-to-end Development, System Design, Secure and Scalable Systems
Tools & Infrastructure: vLLM, Docker, Jenkins, GitHub Actions, Ansible, n8n, Claude, Codex
Projects
- vLLM: Contributed to the open-source vLLM project.
- BuildeerAI: AI-powered engineering app to build things in physics-simulated environments.
- Sikshya: AI-native education platform where teachers can add curriculum for grounded answers, grade papers using AI, and more.
- AI News Agent, Legendary Stories, and others: Automated social media content generator built with n8n and Python.
- AI Stock Picker: Stock selection tool using machine learning and statistics, written in Python with the Alpaca Markets API.
- AI Quant Firm: System that auto-improves financial ML models (running locally on a Mac mini).
Languages
English, Nepali, Hindi, Spanish