Overview

AI Engineer Jobs in Dhaka, Dhaka, Bangladesh at Sodick Co., Ltd.

Title: AI Engineer

Company: Sodick Co., Ltd.

Location: Dhaka, Dhaka, Bangladesh

Sodick America is opening an offshore R&D hub in Bangladesh — our first AI-dedicated research team focused entirely on advancing the intelligence of Sodick's precision industrial machines. Sodick is a world leader in EDM (Electrical Discharge Machining), wire-cutting, and high-precision CNC equipment. This team will push the boundary of what those machines can do through applied AI.

Qualifications

  • Research and develop LLM-based systems for Sodick machine operators — including chatbots, fault diagnosis assistants, and parameter recommendation engines.
  • Surrogate models — build fast, differentiable approximations of EDM physics simulations to accelerate parameter optimization without running costly real-world trials
  • Design and implement RAG (Retrieval-Augmented Generation) pipelines over Sodick technical manuals, maintenance logs, and EDM process knowledge bases.
  • Apply Reinforcement Learning (RL and RLHF) to optimize EDM machining parameters such as pulse energy, gap control, and surface finish quality.
  • Develop Federated Learning frameworks that allow AI models to train across distributed machine fleets at customer sites — without exposing sensitive process data.
  • Build and maintain on-premise AI inference infrastructure — including vLLM serving, quantization (FP8/INT4), and GPU resource management on multi-GPU servers.
  • Design and evaluate RAG quality using structured metrics (e.g., Ragas), gold Q&A benchmarks, and domain-expert feedback loops.
  • Explore and prototype novel AI applications: predictive maintenance, anomaly detection, tool wear estimation, and computer vision for machined part inspection.
  • Collaborate with Sodick Japan and Sodick America engineering teams — document research in English and contribute to internal technical reports and patent proposals.

Upload your CV/resume or any other relevant file. Max. file size: 800 MB.