Open to AI Engineer & Data Scientist roles

Yamin Hossain

AI systems built to work beyond the demo.

I ship AI that earns its keep — zero-LLM heuristic supervisors, hybrid RAG with BM25 reranking, and multimodal clinical pipelines. Latency-obsessed, source-grounded, readable at 2am.

See selected workGet in touch
In the systemsHybrid SQL + vector retrievalTri2DirectZero-LLM heuristic routingFrameIQ6 parallel RAG sub-agentsQurany AI2-layer semantic cacheQurany AI5,700+ titles indexedFrameIQStructured CoT → Pydantic outputNurseLens
— 01 / About

Engineer, scientist, researcher — in that order on Mondays.

04years of focused practice

I build retrieval-grounded AI systems that are honest about what they don't know — and fast enough that you don't notice the work underneath.

I'm an AI Engineer, Data Scientist, and Researcher focused on retrieval-augmented systems, multi-agent architectures, and applied machine learning. I have shipped commercial AI products across travel, healthcare, Islamic education, and B2B matchmaking — and published peer-reviewed work in computer vision and optimal control. I care about source-grounded outputs, latency budgets, and code that other engineers can read at 2am.

# retrieval pipeline I actually ship
retriever = HybridRetriever(
    dense  = ChromaDB(k=20),
    sparse = BM25(k=20),
    rerank = FlashRank(top_k=5),
)
# source-grounded · latency-aware · 2am-readable
5+
Production AI systems shipped
5
Peer-reviewed publications
4
Years of focused practice
AI · DS · Research
Specializations
— 02 / Production Systems

Systems built to stay up, not just to demo well.

Each one links to a case study covering architecture, stack, and what shipped.

03featured of 24 projects
Professional Work · 2026Flagship

Betopia AI Products

A connected AI suite for model access, AI workspaces, platform operations, and AI-assisted software development.

  • Model access — developer platform for model discovery, API keys, and usage analytics
  • Platform operations — user, provider, and limit controls with live health
  • AI workspaces — streamed chat, Deep Research, files, and Knowledge Base context
  • AI-assisted development — terminal CLI and desktop app builder with permission gates
Next.jsTypeScriptWebSocketsRAGAI Developer Tools
Read the case study
Developer Platform · model discovery, API keys, usage analytics, request logs

Developer Platform · model discovery, API keys, usage analytics, request logs

0106
Professional Work · 2026

Tri2Direct AI Travel

Enterprise conversational AI hotel recommendation system built with a dual-layer LangGraph, Django Auth, FastAPI, and Groq (Llama 3).

LangGraphFastAPIDjangoGroq
Case study
Professional Work · 2026

Wiiz AI Agency Suggestion Bot

Real-time AI consultant handling multi-lingual agency matchmaking via WebSockets and Qdrant.

Django ChannelsWebSocketsQdrantLangChain
Case study
— 03 / Experience

Roles, in reverse chronology.

Select any role to explore the details — scope, team photos, and stack.

05positions held
— 04 / AI Assistant

Ask anything about my work.

YHpowered by Gemini

My portfolio assistant knows my full stack, project history, publications, and availability. Responses are grounded in actual data — not hallucinated summaries.

Answers open in the chat panel · Powered by Google Gemini
— 05 / Archive

Everything else, in one place.

24 projects on record. The Betopia suite is presented as one flagship system, so the archive lists the remaining 23.

23projects shown
01Betopia Platform AdminProfessional WorkNext.jsTailwind CSSJWT202602Betopia ChatProfessional WorkNext.jsWebSocketsZustand202603Betopia Chat AdminProfessional WorkNext.jsReactAxios202604Betopia Coder CLIProfessional WorkGoSQLiteLSP202605Betopia Coder DesktopProfessional WorkTauriRustReact202606Tri2Direct AI TravelProfessional WorkLangGraphFastAPIDjango202607Wiiz AI Agency Suggestion BotProfessional WorkDjango ChannelsWebSocketsQdrant202608GopherNotebookAI AgentsGolangWeaviateNext.js202609NurseLens AI BackendProfessional WorkFastAPIOpenAI / xAIPydantic202610Qurany AIProfessional WorkLangGraphRAGFastAPI202511AI Automation SuiteAI Agentsn8nAutomationGemini API202512Reddit Content ClassifierMachine LearningMLOpsNLPReddit API202513NYC Subway RidershipData AnalysisOpenDataPowerBIDashboard202514Cardiac Care PerformanceData AnalysisHealthcareTableauDataViz202515OmnisoundProfessional WorkElevenLabsTavilyStreamlit202516FinanceAIAI AgentsNext.jsTailwindShadcn UI202417Cortex AIAI AgentsLangChainLLMMultiRAG202418FrameIQAI AgentsFlaskLangGraphPostgreSQL202419Art Intelligence HubMachine LearningKerasOpenCVTransfer Learning202420Supply Chain DashboardData AnalysisPythonPandasStreamlit202321Avocado DashboardData AnalysisShinyPythonInteractive202322Netflix InsightsData AnalysisSQLTableauEDA202323HR AnalyticsData AnalysisSQLPower BIHR2023
— 06 / Publications

Peer-reviewed research, written and co-authored.

Computer vision, optimal control, and the labour-market impact of machine learning. Author order shown as published.

05papers in Springer / IEOM
2025

Machine Learning for Computer Vision: A Comprehensive Overview

International Conference on Data Mining and Information Security (ICDMIS 2024) · Springer · pp. 93–104

Focuses on detecting sports types from images for action recognition.

Abdullah Ibnah Hasan, Nayem Uddin Prince, Md. Kamrul Siam, Md. Maruf Miah, Md. Miskat Hossain Siddique, Yamin Hossain, Md. Habibur Rahman
Neural networkImage processingComputer visionFeature extractionMachine learning
Springer
2025

Improving Sports Categorization in WiFi-Blocked Areas: Modifying YOLO Network for All-Inclusive Sports Image Identification

4th International Conference on Machine Learning, IoT and Big Data (ICMIB 2024) · Springer · pp. 229–242

A modified YOLO network for sports image identification in constrained network environments.

Md. Habibur Rahman, Samar Das, Mahtab Uddin, Aminul Islam, Yamin Hossain, Abdullah Hafez Nur, Md. Abdullah
Deep LearningObject detectionYOLOCNN
Springer
2024

YOLOv8 Image Processing for Evaluation of Stability Algorithms Based on Neural Networks: A Sports Use Case

International Conference on Inventive Communication and Computational Technologies (ICICCT 2024) · Springer · pp. 613–622

YOLOv8-based image processing applied to the evaluation of neural-network stability algorithms.

Md. Habibur Rahman, A. S. M. Mohiul Islam, Abdullah Ibnah Hasan, Mahtab Uddin, Ashek Ahmed, Asif Ahammad Miazee, Yamin Hossain
Deep learningYOLOv8Neural networkSports Image Classification
Springer
2024

Optimal Control Problem and Its Application in COVID-19 Transmission Dynamics

International Conference on Advanced Engineering, Technology and Applications (ICAETA 2024) · Springer · pp. 480–493

Optimal control techniques applied to modeling vaccination strategy within COVID-19 transmission dynamics.

Md. Habibur Rahman, Mahtab Uddin, Yamin Hossain, Nijam Uddin Sarkar, Md Aminul Islam, Nafees Siddique, Ashek Ahmed, Shaikh Afnan Birahim
Optimal ControlCOVID-19VaccinationPontryagin Maximum Principle
Springer
2024

The Impact of the Fourth Industrial Revolution and Machine Learning on Employee Skill Sets for Sustainable Survival in the Retail Industry

6th Industrial Engineering and Operations Management Bangladesh Conference (IEOM BD 2023) · pp. 89–112

Explores how employee skill sets in retail are reshaped by the Fourth Industrial Revolution and machine learning.

Md. Habibur Rahman, Md Aminul Islam, Yamin Hossain, Rejwan Bin Sulaiman, Md. Nayeem Chowdhury, Abdullah Hafez Nur
Fourth Industrial RevolutionRetail IndustryEmployee SkillsMachine Learning
IEOM
— 07 / Toolkit

Stack and disciplines I reach for.

4domains of depth
Language & Runtime
Python
Go (Golang)
SQL
R
TypeScript / JS
AI & ML Systems
LangGraph
LangChain
RAG Pipelines
Multi-Agent
Vector DBs
TF · PyTorch · Keras
Computer Vision (YOLOv8)
Data & Analytics
Pandas · NumPy
Plotly · Seaborn
Tableau
Power BI
Streamlit · Shiny
Excel · SQL EDA
Backend & Infra
FastAPI · Django
PostgreSQL · pgvector
Docker
GitHub Actions
Hadoop
Problem SolvingCritical ThinkingCommunicationTeamworkAttention to DetailTime ManagementAdaptabilityContinuous Learning
— 08 / Education & Community

Foundations, coursework, and the things I led on the side.

B.Tech, Computer Science & Engineering
Specialization in Data Analytics · Minor in Mathematics
Vellore Institute of Technology, Andhra Pradesh
2021 — 20258.36 / 10
  • · Data Structures & Algorithms
  • · Probability & Statistics
  • · Linear Algebra
  • · Operating Systems
  • · Object-Oriented Programming
  • · Database Management System
  • · Computer Architecture
  • · Computer Networks
  • · Software Engineering
  • · Optimization Techniques
  • · Data Warehousing & Data Mining
  • · Mining Massive Datasets
  • · NoSQL Database
  • · Application of Data Analytics
Marketing Team Member
LIT-DAC Club
Live Innovate and Transform — Data Analytics
Promoted seminars and events, increasing participation by 40%. Showcased marketing and public-speaking skills, creating an engaging environment for the data-analytics community.
2022 — 2023
Management Team
CSI VIT-AP Chapter
Computer Society of India
Organized tech events and led a team to enhance community service and academic initiatives, fostering unity among 30+ members.
2021 — 2022
— 09 / Certifications

NVIDIA, DataCamp, Google, Hugging Face — and the rest.

27certificates earned
Certificate of Completion: Introduction to Agent Skills
Anthropic
May 2026
Claude Code in Action
Anthropic
Jun 2026
Building with the Claude API
Anthropic
Jun 2026
Multi-Agent Systems with LangGraph
DataCamp
2025
LLMOps Concepts
DataCamp
2025
AI Agents Fundamentals
Hugging Face
2025
AI-Driven Biomedical Signal Analysis: ECG and EEG Processing for Healthcare Applications
VIT-AP University
2025
Complete Research Training: From Zero to Publication
Bangladesh Advanced Research Training Center
2025
— 10 / Contact

Let's build something worth shipping.

Email
issan.yamin@gmail.com
Location
Bangladesh — Open to remote
Looking for
AI Engineer · Data Scientist roles, research collabs
Response time
Usually within 24 hours