SHOUMIK SARKAR

AI/ML Engineer

Dhaka, BD.

About

Highly skilled AI/ML Engineer specializing in multimodal learning, computer vision, and production-oriented ML systems, with expertise in developing end-to-end pipelines using PyTorch, FastAPI, and transformer-based architectures. Proven ability to deliver high-accuracy solutions across vision, audio, and NLP tasks, demonstrating a strong foundation in scalable AI infrastructure and inference system deployment. Eager to apply advanced machine learning techniques to drive innovation and achieve measurable results in dynamic environments.

Projects

DeceptiLens: Multimodal Deception Detection

→

Summary

Developed and deployed an end-to-end multimodal deception detection pipeline, integrating facial landmarks, audio descriptors, and deep visual embeddings for real-time inference using PyTorch, OpenCV, Librosa, MediaPipe, FastAPI, and Streamlit.

RumorFusionX: Multimodal Fake Social Media Post Detection

→

Summary

Engineered a multimodal misinformation detection pipeline for social media posts, leveraging OCR, image features, and LLMs to achieve high classification accuracy using PyTorch, OCR, Transformers, OpenCV, and FastAPI.

MedHelp: Medical Information Assistant

→

Summary

Created a medical-domain conversational AI assistant utilizing transformer-based models and retrieval-augmented generation, enhancing information retrieval and comparative search capabilities with FastAPI and PostgreSQL.

Education

North South University

Dhaka, Dhaka, Bangladesh
→
BSc Computer Science and Engineering Grade: 3.2 out of 4.0

Courses

Intelligent System Engineering

Skills

Programming Languages

Python, C++, JavaScript, SQL.

Machine Learning

PyTorch, TensorFlow, Scikit-learn, Transformer, OpenCV, Hugging Face.

AI Domains

Multimodal Learning, Computer Vision, NLP, RAG, OCR, LLM Fine-Tuning.

Backend & Deployment

FastAPI, PostgreSQL, Supabase, REST APIs.

Data & Tools

MS Office, Pandas, NumPy, MySQL, Git, GitHub Actions, VS Code, Linux.

Soft Skills

Adaptability, Analytical Problem Solving, Team Collaboration, Research & Documentation.

Certificates

Vision Language Models (VLM) Bootcamp

OpenCV University