Applied AI & Interactive Systems Researcher

Building real-world AI and interactive systems with a focus on agentic workflows and human–AI interaction. My work involves designing and evaluating deployed AI tools and interactive systems, using behavioral data and controlled experiments to study user understanding, trust, and decision-making.

GitHubOSF

Research Interests

Human-AI Interaction

Investigating mechanisms for trust and clarity in collaborative loops between humans and autonomous agents.

Agentic AI

Developing evaluation-driven agent workflows for multi-step reasoning and complex tool interaction.

Interactive/XR Systems

Building experimental platforms that leverage spatial computing for new forms of human-computer synergy.

Behavioral Data

User-centric evaluation through the analysis of interaction logs and physiological behavioral signals.

Research

COMPLETED2025

Medical Reports AI Assistant

Designed and evaluated an AI system for medical report understanding using Python, LangGraph, OCR, and retrieval-augmented generation. Assessed effects on user clarity, confidence, and trust.

IN PROGRESS

Duration-Weighted Disfluency & L2 Speech Fluency

Investigates whether a duration-weighted disfluency metric aligns more closely with human fluency judgments than traditional rate- or count-based fluency measures in L2 speech. The study focuses on perceptual fluency and does not evaluate overall speaking proficiency.

Applied Projects

Speech Analysis API (IELTS-Aligned)

English proficiency assessment system

Research Motivation

Determining if automated speech analysis can provide rapid, reliable assessment of English fluency and pronunciation.

System Built

Built with Python, Whisper, Wav2Vec and FastAPI. Provides band-style scores across fluency, pronunciation, vocabulary, and grammar dimensions.

Evaluation / Data

Deployed as production SaaS on RapidAPI

Key Insight

"Automated assessment alignment with IELTS standards enables rapid proficiency evaluation without human raters."

Video

Medical Reports AI Workbench

Upload, manage, and chat with reports using AI

Research Motivation

Building an interactive workbench for medical professionals to upload, organize, and query reports through a conversational AI interface.

System Built

AI-powered platform with document upload (PDFs, images), OCR processing, semantic embeddings for searchability. Uses LangGraph for multi-step reasoning, LangChain for chains, and OpenAI (GPT-4o mini) for chat. UI built with Gradio. Storage: Backblaze S3 for files, Supabase (Postgres + vector search) for metadata and embeddings.

Evaluation / Data

Deployed as interactive workbench. Supports full-screen chat interface for real-time report querying with document context.

Key Insight

"Combining OCR, semantic search, and agentic AI enables seamless document-aware chat for professional workflows."

Video

VR Projects

Immersive experiences with spatial input

Research Motivation

Exploring immersive VR experiences that combine audio-reactive environments and real-world mechanical input for novel interaction paradigms.

System Built

Created a music VR project with audio reactivity. Co-created a surfing VR game that accepted input from a mechanical surfboard, translating physical movement into in-game navigation and actions.

Evaluation / Data

Demonstrated proof-of-concept for peripheral device integration with VR and validated immersive gameplay through playtest feedback

Key Insight

"Mechanical input devices enhance immersion by creating direct physical-to-digital feedback loops, improving user engagement in VR experiences."

Video

Video

Applied Educational Game Systems

Learning & Behavior Research Through Interactive Design

Research Motivation

Designing interactive learning systems that explore how mechanics, constraints, and feedback loops influence learning, decision-making, and retention.

System Built

Implemented four game prototypes—Grammar Sorts, Reverse Math, Crave Control, and Jazzy's Sight Word Jumble—each targeting specific learning dimensions: concept formation, problem-solving strategies, behavior change, and multimodal learning.

Evaluation / Data

Deployed and iterated based on player feedback and learning outcomes. Each system investigates specific learning mechanics through interactive prototypes and user testing.

Key Insight

"Game design mechanics directly influence learning outcomes: feedback timing shapes strategy, difficulty progression affects cognitive load, and choice architecture guides behavioral decisions."

Grammar Sorts Demo

AI Book Translator

Multi-language document translation system

Research Motivation

Designing semantic translation pipelines with quality validation across multiple languages.

System Built

Processes PDFs page-by-page, uses Claude API for semantic translation, and OpenAI as evaluator for quality validation. Generates four DOCX outputs in different languages while preserving page structure.

Evaluation / Data

Maintains document structure and numbering across all output languages with semantic accuracy validation

Key Insight

"Multi-stage translation pipeline with AI-driven quality assurance enables rapid document localization at scale."

Video

WhatsApp AI Customer Support System

Automated customer service chatbot

Research Motivation

Testing whether conversational AI can automate customer support with intelligent issue categorization and response generation.

System Built

AI-powered WhatsApp chatbot that converses with customers, collects complaints, and uses OpenAI to categorize issues automatically using n8n workflow automation.

Evaluation / Data

Deployed for real customer support with real-time categorization metrics

Key Insight

"AI-driven categorization reduces manual triage time while maintaining customer satisfaction through natural conversation."

Video

Data-Driven Evaluation of Player Behavior in a Mobile Shooter

Mobile game - 100k+ downloads

Research Motivation

Investigating how data-driven gameplay design and A/B testing can optimize player retention and monetization.

System Built

Mobile cover-shooter game with analytics integration for gameplay metrics, A/B testing framework, and real-time balance iteration.

Evaluation / Data

FTUE optimization, retention metrics (D1/D7/D14), engagement and monetization analysis across player cohorts

Key Insight

"Analytics-driven game design improved D1 retention by 10% and Playtime by 22% through iterative feature balancing."

Video

Methods & Technical Experience

AI AND AGENTIC SYSTEMS

  • OpenAI API
  • LangGraph / LangChain
  • AutoGen
  • Claude API
  • RAG pipelines
  • n8n Automation

EVALUATION & BEHAVIORAL ANALYSIS

  • A/B Testing
  • Retention Analytics (D1-D14)
  • Statistical Analysis
  • Behavioral Coding & Metrics

INTERACTIVE SYSTEMS & GAME DEV

  • Unity (C#)
  • Gameplay Architecture
  • GameAnalytics
  • Human-Computer Interaction

BACKEND & DATA SYSTEMS

  • FastAPI
  • Next.js
  • Supabase
  • Supabase VectorDB
  • BackBlaze B2

Academic Background

B.Sc. in Computer Science & Cognitive Systems

Islamic University of Technology, Bangladesh

2015 – 2018

CGPA: 3.57 / 4.00

Relevant Coursework: Human-Computer Interaction, Machine Learning, Computer Graphics & Multimedia Systems, Artificial Intelligence, Pattern Recognition, Cognitive Psychology, Computer Vision, Statistical & Mathematical Analysis

Professional Experience

Digital Games Developer

Gameloops — Dhaka, Bangladesh

Aug 2022 – Present

Led development of mobile games and interactive systems from prototyping to release. Designed and evaluated gameplay systems using controlled A/B testing on FTUE, economy balance, and feature selection. Analyzed player behavior using retention (D1/D7/D14), engagement, and monetization metrics to guide iteration. Built and evaluated agentic AI workflows for automation and internal tooling.

Digital Games Developer

Lightneer — Helsinki, Finland

Aug 2020 – Jul 2022

Built and iterated gameplay prototypes using player testing and UX refinement.

Digital Games Developer

KobGames Studios — Tel Aviv, Israel

Jan 2020 – Jun 2020

Designed and engineered gameplay systems across multiple prototypes.

Digital Games Developer

Alienide Interactive — Dhaka, Bangladesh

Sep 2018 – Dec 2019

Developed mobile applications, games, and AR/VR learning tools.