Senior AI Engineer · Recommendation Systems · Applied Deep Learning Senior Software Engineer @ Foodi Express Limited Former Senior Software Engineer @ BJIT Limited B.Sc. in CSE @ CUET · MSc in Data Science @ UIU (Ongoing)
Practical notes on building AI systems that hold up in production: recommendation engines, forecasting, multi-agent workflows, and the engineering decisions in between.
The first articles are on their way. Check back soon.
I am Md Aminul Haque Palash, a Senior AI Engineer based in Dhaka, Bangladesh. I build production machine learning systems across recommendation, forecasting, conversational AI, edge inference, and multi-agent applications.
I currently work at Foodi Express Limited, where I develop personalized food and branch recommendations, delivery-time estimation, demand forecasting, and scalable data infrastructure for a large food delivery platform.
Previously, I spent more than six years at BJIT Limited building AI products for travel, food, mobile, and conversational experiences. My work included a LangGraph-based travel assistant, an on-device recommendation system, and an AI travel diary for iOS.
I am open to conversations about applied AI, recommendation systems, production machine learning, and research collaborations.
Email: aminulpalash506@gmail.com
Phone: +8801960512706
GitHub: github.com/aminul-palash
LinkedIn: linkedin.com/in/aminulpalash
Dhaka, Bangladesh
Building personalized food and branch recommendations, DeepETA delivery estimation, demand forecasting, scalable data pipelines, and low-latency APIs for a large food delivery platform.
Dhaka, Bangladesh
Designed and shipped AI products across travel, food, mobile, and conversational experiences, including a multi-agent travel assistant, personalized suggestions, edge recommendation systems, and a Rasa chatbot.
Pioneer Alpha Ltd. · Dhaka, Bangladesh
Contributed to applied AI research and supported the development of machine learning solutions as a volunteer research lead. This work included Bangla NLP, clinical AI, speech AI, computer vision, and research publications across image captioning, text-to-image generation, graph neural networks, plant disease detection, and brain tumor segmentation.
Dhaka, Bangladesh
Mentored engineers in deep learning, RNNs, LSTMs, GRUs, Transformers, NLP, generative AI, and responsible AI practice.
Foodi is a US-Bangla Airlines concern and an online food delivery platform operating across Bangladesh, connecting customers with more than 5,000 restaurants and 2,000 riders.
As a Senior Software Engineer, I work with millions of users and millions of daily interaction data points across food delivery and Foodimart grocery experiences, turning that scale into better discovery, forecasting, and delivery decisions.
Five peer-reviewed publications across Bangla NLP, computer vision, medical AI, and generative models, with 140+ Google Scholar citations, an h-index of 5, and an i10-index of 5.
From multi-agent travel assistants and edge AI to production APIs and scalable pipelines, I also mentor engineers in deep learning, Transformers, NLP, generative AI, and responsible AI.
I enjoy understanding how technical systems meet real people and real constraints. That curiosity leads me toward product thinking, system design, research reading, and mentoring engineers.
Outside day-to-day engineering, I make time for:
The best work, for me, sits where deep technical craft meets useful human outcomes.
Built recommendation systems for menu items and branch suggestions using millions of order and user interaction records, improving relevance and discovery across the platform.
Developing food preparation and delivery-time estimation models using historical order trends, restaurant workload, and peak-hour traffic. Also designing demand forecasting for inventory and rider planning.
Designed a modular LangGraph and LangChain system for personalized travel planning, integrating search, routing, and weather APIs for dynamic recommendations.
Engineered an offline Android and iOS recommendation system with ONNX Runtime and RNN-based sequential models. Achieved 88% food-type prediction accuracy and an 81% Top-5 recommendation hit rate.
Built an iOS application that generates personalized travel diaries from activity data, handling more than 10,000 daily background events and reducing manual input by around 90% in internal testing.
Worked on a Rasa-based assistant with intent classification, entity extraction, dialogue management, REST APIs, and context-aware responses.