Open to opportunities

Soumya Dahal

@soumyadahal1

Message

I build computer vision and multimodal ML systems from training through API deployment.

Nepal

Message

What I'm looking for

I'm looking to build computer vision and multimodal ML products, contributing across data pipelines, model training, evaluation, and deployment while growing through open-source and production engineering work.

I've built image captioning systems using EfficientNet, Vision Transformers, Transformer encoder-decoder architectures, and GPT-2-style decoders. My work spans COCO 2017 training pipelines, SCST optimization, BLEU-4 and CIDEr evaluation, Hugging Face publishing, and FastAPI inference services.

At Shaman Yantra, I built backend services and a multi-tenant Business Management System with inventory, CRM, reporting, and Meta Messenger chatbot integration. I also develop real-time ML applications with Python, Go, React, and microservice-based architectures.

Experience

Work history, roles, and key accomplishments

Software Engineer

Shaman Yantra

Jan 2026 - Mar 2026 (2 months)

Designed and built a multi-tenant Business Management System (BMS) with inventory, CRM, and reporting modules. Integrated an AI chatbot with Meta Messenger and developed backend services for message ingestion and routing.

Python Go JavaScript fastAPI React REST APIs Chatbot