Computer Vision
Developing visual models that extract robust signals from images and support evidence-led decisions.
AI / ML Engineer Open to opportunities
Buildingintelligentsystemsthat notice.
Turning signals, pixels, language, and data into useful AI experiences.
Iterate, Curate, Adapt
Scroll to explore
02 / ABOUT
About Shoumik
I build intelligent systems that understand, learn, and solve real-world problems.
Based inDhaka, Bangladesh
Available forInternship and full-time opportunities - available to start immediately.
What I do
Language & Multimodal AI
Combining OCR, language models, and visual context to understand information beyond a single modality.
Audio Intelligence
Working with acoustic descriptors and audio-visual fusion for richer, context-aware AI systems.
AI Product Systems
Taking ideas from research and data pipelines to usable APIs, interactive tools, and polished interfaces.
My approach
I start by understanding the problem, data, and people involved. Then I build iteratively - from research and prototyping through evaluation, deployment, and refinement - with a focus on responsible, practical outcomes.
Currently
Machine Learning Engineer Intern
Flyrank AI
Contributing to end-to-end ML pipelines and AI-driven products, with hands-on work in dataset evaluation, Python-based analysis, and scalable deployment workflows.
03 / PROJECTS & RESEARCHSystems that
Systems that
see differently.
PROJECT / 01
DeceptiLens
An end-to-end multimodal deception-detection experience that combines face, audio, and visual evidence into a single prediction. Built for transparent, real-time exploration of complex behavioural signals.
- Built an end-to-end pipeline combining facial landmarks, audio descriptors, and deep visual embeddings.
- Fused MediaPipe, Librosa, and ResNet-18 features with a custom cross-attention model.
- Achieved 82.11% accuracy on the DOLOS benchmark.
- Deployed an interactive Streamlit app with real-time performance visualisations.
04 / SYSTEM ATLASA flexible
A flexible
toolkit.
Tap a node to see how I use it. These are the tools I reach for when an idea needs to become a reliable system.
AI
systems
thinking
systems
thinking
PyTorch
Custom models, cross-attention, experiments, and practical training workflows.
Custom models, cross-attention, experiments, and practical training workflows.
05 / EXPERIENCELearning in
Learning in
public.
Machine Learning
Engineer (Intern)
Flyrank AIContributing to the design and development of end-to-end ML pipelines and AI-driven digital products, with a focus on process efficiency and responsible deployment. Working alongside senior engineers to translate system requirements into technical solutions for local, scalable AI tools, while supporting dataset evaluation and performance analysis through Python-based data workflows.
Research
Engineer
VerdantConducting requirement research and technical analysis for client software products before development, translating product needs into actionable specifications. Contributing across project-based software builds, including API integration and database management, while collaborating with the development team to scope work and keep deliverables aligned with client requirements.
06 / EDUCATIONA foundation for
A foundation for
useful systems.
BSc, Computer Science & Engineering
Relevant Coursework
Data Structures & AlgorithmsTheory of ComputationProgramming Languages (C, C++, Java)Machine LearningDeep LearningPattern Recognition & Neural NetworksComputer Vision
Vision Language Models Bootcamp
OpenCV University, 2026. An intensive exploration of multimodal AI architectures, visual feature manipulation, and vision-language model workflows.