Aaditya Chapagain
Tech Lead, AI Team at Accelo
Seven years across machine learning, data platforms and AI engineering — Kathmandu to Wollongong. I build agent systems and the observability that keeps them honest, and I care most about the part where a system admits what it does not know.

WHAT I HAVE STUDIED SO FAR
Education.
Master of Computer Science: Machine Learning and Big Data
- Specialisation in machine learning and big data
- Graded 80%
- Studied alongside full-time consulting work
Bachelor of Electronics and Communication Engineering
- Focus on embedded systems and signal processing
- Signal processing groundwork that later fed into speech recognition work
MY PROFESSIONAL JOURNEY
Work Experience
Senior AI/ML Engineer → Tech Lead, AI Team
- Promoted to Tech Lead for the AI team, owning end-to-end delivery of production AI products
- Architected and shipped a production agentic AI platform on LangGraph and AWS Bedrock, with multi-agent workflows, streaming, and a human-in-the-loop system
- Built Nova, a customer-facing AI chat agent backed by an MCP server with 90+ tools, enabling safe, auditable querying and updating of platform data
- Designed AI observability and evaluation covering agent runs, LLM calls, tools, tokens and cost, with LLM-as-a-judge evaluations to improve quality and reliability
- Led engineering decisions on long-running agents, token limits, prompt caching, model selection, performance and cost; built supporting pipelines with Kinesis, S3, DuckDB, ClickHouse and Athena
- Used Claude Code extensively for AI-assisted development across architecture, agents, MCP services, data pipelines, testing and deployment
Senior Machine Learning and Data Consultant
- Developed advanced data models and algorithms using dbt, Snowflake and AWS, enabling data warehousing and predictive analytics over large-scale clickstream data
- Led complex data pipeline migrations to improve scalability and performance for major energy retail companies
- Architected and deployed scalable data platforms on AWS using Terraform, supporting real-time analytics
- Implemented data quality checks and profiling processes to maintain data integrity and governance
- Optimised ETL/ELT processes, improving execution efficiency and transformation accuracy for warehouse management systems
- Implemented Mage-AI on AWS ECS, creating hundreds of data pipelines integrated with Apache Iceberg on AWS and Snowflake
Software Engineer
- Researched and implemented machine learning and data science techniques for information extraction and classification of digital documents
- Trained and deployed Large Language Models for information extraction
- Designed REST APIs and developed workflows on AWS and Azure, integrating data science solutions into business operations
Machine Learning Engineer
- Engineered and deployed end-to-end ML pipelines for NLP and computer vision, using PyTorch and TensorFlow for model development and optimisation
- Built a production recommendation system on GCP using Neural Collaborative Filtering to match users with food products from preference patterns
- Developed scalable ML services with FastAPI, containerised with Docker
- Trained and deployed a Hugging Face question answering model to extract key summaries from speech and audio-based questionnaires
- Built a child-safe chatbot using Large Language Models, trained on an HPC cluster across 40 GPUs
Personal Projects
Explore a collection of my recent projects showcasing my skills in software development, machine learning, and data science. Each project represents my passion for building innovative solutions to real-world problems.
Markdown To PDF Generator
Online app to generate pdf from your markdown.
Technologies
Highlights
- •Implemented custom markdown to pdf converter using remark
- •Several theme of pdf can be generated
- •can process markdown file
Online Code Editor
Interactive Online code editor with autocomplete
Technologies
Highlights
- •Responsive and draggable code editor
- •code autcomplete for js, python, rust
Financial Calculator
Visual tool to estimate your investment on various sectors and analyzing loans to better manage your financial
Technologies
Highlights
- •Generate charts of of future Financial status on varoius Terms
- •Compare various Investments and Loans scheme with different variables
Loan Optimizer
Tool to help you decide how to optimize your loan structure to make your life easy
Technologies
Highlights
- •Help you with your loans, if you have any to make impactful decisions
Certifications
Databricks Certified Data Engineer Associate
Databricks
2023
- Data pipeline development
- ETL and ELT processes
- Distributed computing
- Performance optimisation
TensorFlow Developer Certificate
2022
- Neural network implementation
- Computer vision applications
- Natural language processing
- Deploying TensorFlow models
Natural Language Processing
Coursera
2022
- Sentiment analysis
- Language modelling
- Neural machine translation
- Transformers and BERT
Machine Learning and Deep Learning
Coursera
2021
- Supervised and unsupervised learning
- Deep neural networks
- Model evaluation and tuning
TECHNICAL PROFICIENCIES
Skills.
<programming>
</programming>
<itTools>
</itTools>
<database>
</database>
MY INSIGHTS & THOUGHTS
Blogs.
Guide to Data Structure in pythonverse
Datastructures are essential building blocks upon which any programs are constructed. Each data-structure offers a unique method of arranging data, So It can be retrieaved effectively. Python's Standard Library contains a large collections of data structures.
Which would I choose Rust or Go ? `Caught between a rock and a hard place`
Rust for developing systems and performance-critical applications, Go for implicity and scalability of mircro-services and distributed systems.
Understanding useEffect in React - A Beginner's Guide
UseEffect hook is a powerful and versatile tool in react that allows you to handle side effects in your components.
GET IN TOUCH
Contact.
Location
Wollongong, NSW, 2500
Australia