Job Experience
Leading a team of AI engineers designing and deploying LLM-driven solutions for client projects, across AI-powered marketplace catalog operations — from solution architecture and experimentation through to production deployment.
Architected an AI-based content fidelity platform for an industry-leading US e-commerce aggregator managing 10K+ SKUs, built with Python, Django, AWS Bedrock, Docker, Kubernetes and GitLab CI. It keeps product data accurate across live marketplace listings, improving monitoring speed by 200% and reducing manual effort by 80%.
Architected an agentic competitor content research and trend adoption platform using AWS Bedrock, Amazon EKS, Google Vertex AI, CLIP, DINOv2, YOLO, Python and Django. It identifies winning marketplace content patterns and translates them into listing optimization strategies, demonstrating a 17% projected gross revenue uplift during A/B testing.
Developed large-scale data pipelines using Python, Django, Playwright, Celery, Bright Data, Snowflake and PostgreSQL, improving processing reliability, scalability and automation.
I also own end-to-end client communication, technical direction and stakeholder alignment for large-scale, company-wide AI adoption initiatives.
Led Back-End development on IQVIA's Orchestrated Analytics and GenAI products, coordinating a team of Python/AI developers to build an AI-based Chat Assistant and improve system performance and code quality. Led the integration of the OA Analytics platform with a client's MS Teams app using Python, FastAPI and the MS Graph API, improving notification delivery performance by 50%. Also architected a centralized, event-driven task scheduler with Celery Beat and FastAPI that reduced task duplication by 90%.
Led AI adoption and mentoring initiatives to keep teammates current with AI trends, industry best practices and tooling — MCP, RAG systems, advanced prompt engineering and eval systems.
Implemented the Predictive Analysis for Chemist shops for Pharma business of ACI. It's a forecast generation process to enable the Pharma field force to plan better with the help of purchase predictions buying patterns provided by AI algorithms.
Also implemented an AI-driven chatbot for the consumer brand customers with an engaging experience and precise information to guide them to the proper channel. It also helps ACI consumer brands to never miss any business opportunities due to unavoidable service outage from human agents.
Implemented a human movement detection service for retail stores using Python, PyTorch and YOLO, achieving detection accuracy up to 80%.
Designed scalable ETL pipelines using Python, Django and SQL Server Agent for embedded Power BI dashboards, improving service uptime and dynamic report availability by 50%.
Implemented a tractor movement tracking system for ACI Motors, building an IoT server back-end with Django, Node.js and Socket.io to handle 10k+ GPS devices for tractors.
Guided a team of AI professionals to solve the complex problems of ACI businesses. Also worked with ACI business professionals from different fields to help them adapt to data-driven technologies to improve KPI.
Worked with user behavior data from Bangladesh's one of the biggest retail chains — Shwapno, owned by ACI. Coordinated a small team of AI developers and analysts to perform customer segmentation, churn analysis, basket analysis and purchase prediction.
Implemented the Back-End APIs & server automation of the Teamwork platform — a field-force tracking and task management solution including web and mobile platforms. Provided APIs for GPS tracking, attendance management, reporting, task management, reminders, chatting, file sharing, access control and assignment management. Coordinated a team of mobile app developers, UX designers and front-end developers.
Built the Dingi Vehicle tracking system from the ground up and helped onboard its first enterprise clients — implementing an IoT server back-end using Django, Node.js and Socket.io to handle 50k+ GPS/OBD devices for clients like GrameenPhone.
Improved the performance of the Dingi Map API service by 150% and reduced AWS cost by 50% by optimizing Python multiprocessing, Django, AWS elastic load balancers and autoscaling. It's used by ride-sharing companies and vehicle-tracking providers in Bangladesh.
Improved data-pipeline efficiency for the Dingi Life app by 50% and data accuracy by 80% using PostgreSQL, Elasticsearch and Python.
My first job. Landed this programming-based role straight out of an EEE degree on sheer interest and self-built portfolio projects rather than a formal CS background. It gave me my first real exposure to surveillance system development and IoT projects.
Implemented an intelligent video surveillance system using C#, Emgu CV, OpenCV, YOLO and Python for people-movement detection and tracking, achieving detection accuracy up to 70%.
Also built a PDF-to-image service for document management using C# and ImageMagick, improving document-processing efficiency by 50%.