About Me

Sadman Kabir Soumik

Hello, I am Soumik, pronounced Shou mik.

I am a Senior Data Scientist at Optimizely Inc.. I build production artificial intelligence (AI) systems that go beyond training models. My work includes multi agent systems, large language model (LLM) applications, retrieval augmented generation (RAG) systems, recommender systems, evaluation pipelines, and machine learning infrastructure.

I have seven years of industry experience in artificial intelligence and machine learning. I enjoy the engineering around a model, including memory, application programming interfaces (APIs), data pipelines, deployment, monitoring, and the product decisions that make a system useful and dependable.

My work today

Some of the systems I work on at Optimizely include:

  1. Multi purpose agentic AI systems and memory components that preserve user and chat context over long periods.
  2. Large scale recommendation systems that serve millions of user interactions every day.
  3. A/B testing systems for evaluating and deploying machine learning models.
  4. Real time training pipelines that learn from user interactions.
  5. Machine learning pipelines that train thousands of models each day.
  6. Conversational AI systems for customer segmentation and analysis.
  7. Unified API services for different large language model providers.
  8. Retrieval augmented generation systems that give AI applications domain specific knowledge.

Experience

  1. Senior Data Scientist at Optimizely Inc., January 2023 to present. I work remotely with teams building AI and machine learning products for a digital experience platform.
  2. Senior Artificial Intelligence Engineer at Venturas Ltd., June 2021 to December 2023. I designed AI and machine learning systems, guided development teams, and worked as a Tech Lead for a period.
  3. Machine Learning Engineer at Chowa Giken Corporation, September 2019 to June 2021. This was the start of my professional career after I completed my bachelor degree.

Education and location

I completed a Bachelor of Science in Computer Science and Engineering at North South University in Dhaka from 2015 to 2019.

I am based in Dhaka, Bangladesh, and open to relocation.

Things I build outside work

I enjoy taking a product from an idea to a working product. I make the product decisions, design the architecture, write the code, deploy the product, and improve it based on real user feedback.

  1. bdtechjobs.com, a curated job board for technology opportunities in Bangladesh.
  2. Jot, a private budget tracker for iPhone, iPad, and Apple Watch. It has no account, no advertisements, and no bank connection. Data stays on your devices, with optional iCloud sync.
  3. Voice Writter, an on device voice to text app for macOS with automatic grammar correction. It uses Whisper and a local MLX language model.
  4. Cycle, a private period tracker for iOS. It has no account, no analytics, and no third party software development kits.

Tools and technologies

I work with Python, TypeScript, SQL, Swift, C, C++, LangGraph, LangChain, the OpenAI software development kit (SDK), Hugging Face, PyTorch, Keras, scikit learn, XGBoost, Pandas, Polars, Apache Spark, Google Cloud Platform, Amazon Web Services, Docker, Kubernetes, Terraform, Airflow, MLflow, FastAPI, Django, Next.js, React, SwiftUI, UIKit, and SwiftData.

I also work with embeddings, retrieval augmented generation, model context protocol (MCP), and tools such as Claude Code, Cursor, and Codex.

Writing and contact

I write on this blog about machine learning systems, artificial intelligence, software engineering, infrastructure, and the lessons I learn while building and operating products.

You can visit my portfolio, connect with me on LinkedIn, or see my work on GitHub.

Email: sadmanks@gmail.com

Disclaimer

This blog is a place for me to share my thoughts and learnings from daily life and work. These are my personal notes. I do not claim to be an expert in every topic I write about. I am a learner who likes to share what I learn.

My notes may not be complete or fully accurate because they are not formal research. They reflect my current understanding and can be used as a starting point for further study. Please read them with an open mind. I am always learning, and I may make mistakes.

Thank you for visiting my site. If you have any questions or concerns, please feel free to contact me.