From research and evaluation to cloud deployment, I work across the full lifecycle of intelligent products.
01
AI/ML Researcher/Engineer
NSDevil Co., Ltd.
Aug 2026 — Present
South Korea
Developing an AI-based automated scoring system for Korean schools across Korean and English subjects.
Running controlled experiments with Quadratic Weighted Kappa (QWK) to improve prediction consistency and scoring reliability.
Performing rubric-level error analysis to identify failure patterns and guide iterative improvements.
AI EvaluationQWKError Analysis
02
AI/ML Engineer
MIAM Technology Pvt. Ltd.
Apr 2026 — Jul 2026
Kathmandu, Nepal
Enhanced RoamX restaurant recommendations with collaborative, content-based, and geolocation-aware personalization.
Provisioned Proxmox VMs as self-hosted inference servers and deployed Qwen2.5 1.5B with llama.cpp for local inference.
Integrated a Hermes-based employee chatbot with Slack for internal support and knowledge retrieval.
Recommendation SystemsProxmoxQwenSlack
03
Machine Learning Engineer (Freelancer)
Self-Employed
Nov 2025 — Apr 2026
Remote
Built agentic AI and RAG solutions with LLMs, LangChain, FastAPI, semantic search, and context-aware generation.
Produced technical design documents and applied MLflow and DVC for reproducible experiments and model versioning.
RAGAgentic AIFastAPIMLflowDVC
04
AI Intern
Fusemachines
Oct 2025 — Jan 2026
Kathmandu, Nepal
Building QPool, an agentic AI platform for generating MCQs, paraphrases, and distractors with a hybrid multi-agent architecture.
Designing multi-stage generation pipelines with prompt engineering, structured outputs, schema validation, formula evaluation, and difficulty-aware processing.
Integrating automated quality checks to reduce malformed LLM responses and improve consistency across diverse question formats.
Exposing the platform through FastAPI REST APIs and contributing through unit tests, code reviews, and maintainable engineering practices.
QPoolMulti-AgentFastAPIPrompt Engineering
05
Fusemachines AI Fellowship
Fusemachines
Apr 2025 — Dec 2025
Remote
Completed an intensive AI training program covering data preprocessing, EDA, feature engineering, classical machine learning, deep learning, computer vision, NLP, LLM fine-tuning, RAG, agentic AI, evaluation, and MLOps.
Completed weekly practical assignments across supervised and unsupervised learning, neural networks, and modern generative AI workflows.
Built an AI Job Recommender and Career Coach as a capstone project, combining hybrid retrieval, embeddings, and personalized recommendations.