I design, build, and deploy production-grade LLM applications, multi-agent workflows, and scalable MLOps architectures that bridge the gap between research and real-world engineering.
🌐 Portfolio | 💼 LinkedIn | 📧 Email | 🎥 YouTube Channel
- Agentic Systems & Copilots: Engineering stateful, multi-agent architectures (using LangGraph, CrewAI, and OpenAI Agent SDK) that automate complex enterprise and clinical workflows.
- Advanced RAG Pipelines: Architecting robust Retrieval-Augmented Generation pipelines integrated with vector databases (ChromaDB, Milvus, Pinecone) and layered security for explainable data analytics.
- Model Optimization: Fine-tuning open-source LLMs (like Gemma) using QLoRA/PEFT and applying quantization techniques to optimize cost, latency, and domain accuracy.
- Production MLOps: Building end-to-end CI/CD infrastructure, containerized deployments (Docker, Kubernetes), and cloud orchestration on AWS (SageMaker, EC2, Lambda, S3).
- +10% Accuracy Lift: Developed an esports prediction model utilizing rigorous statistical analysis that outpaced traditional betting odds.
- +20% User Engagement: Optimized user interaction metrics through a fine-tuned open-source chatbot deployment.
- -35% OCR Errors: Significantly upgraded data extraction reliability via targeted multimodal LLM improvements.
- -20% Simulated Risk: Engineered predictive network logic to minimize operational vulnerabilities.
Nov 2025 – Present | Dhaka, Bangladesh
- Architecting multi-agent clinical systems using LangGraph and OpenAI Agent SDK to accelerate healthcare data analytics.
- Developing secure, high-performance backend APIs with FastAPI to orchestrate complex agent workflows.
Jan 2025 – Oct 2025 | Dhaka, Bangladesh
- Designed autonomous agentic setups for e-commerce automation and multi-agent systems for drug discovery using CrewAI and ChromaDB.
- Managed full application lifecycles using Docker, Kubernetes, and automated CI/CD pipelines on AWS.
July 2023 – Dec 2024 | Remote
- Built AI agents to parse millions of scientific papers, fine-tuned Gemma models via QLoRA, and built secure on-premises search engines for pharmacokinetics data.
| Category | Technologies |
|---|---|
| Generative AI | LLM Fine-tuning (LoRA/PEFT, Quantization), LangGraph, CrewAI, LangChain, Prompt Engineering |
| Vector DBs | ChromaDB, Milvus, Pinecone |
| Core Frameworks | Python, FastAPI, Django, Flask, PyTorch, TensorFlow, Scikit-learn, Pandas |
| MLOps & DevOps | Docker, Kubernetes, GitHub Actions, AWS (SageMaker, EC2, S3, Lambda), MLflow, DVC |
| Databases | PostgreSQL, MongoDB, Redis |
- Data Discovery (YouTube): I run a technical channel sharing practical tutorials on Generative AI architectures, RAG systems, and paper reviews.
- AI Instructor: Delivered basic-to-advanced GenAI curriculums (Transformers, Vector DBs, Agentic Frameworks) at Horizon Solutions and taught AI fundamentals at Skill Hunt.
- Ongoing Research: Focused on Transfer Learning for Bone Marrow Classification and Thermal Comfort Analysis using Explainable AI (XAI).
- BSc. in Information and Communication Engineering – East West University (GPA: 3.46/4.0)