Nasim Mahmud Nayan

Building Production AI Systems That Scale

AI Engineer with 3+ years building production AI systems that drive measurable impact. Specialized in cost-efficient GenAI solutions (LLMs, RAG, AI agents), from wellness platforms serving 100+ daily users to RAG-powered sales agents handling 500+ inquiries/day.

Daily Active Users

1.5k+

Cost Reduction

94%

Students Served

10K+

Experience

3+ yrs

Technical Skills

My proficiency across modern programming languages, machine learning frameworks, and cloud technologies.

Programming Languages

Python

Expert

JavaScript

Advanced

SQL

Advanced

ML Frameworks & Libraries

PyTorch / TensorFlow

Expert

LangChain / LLMs

Advanced

Scikit-learn / XGBoost

Expert

Cloud & DevOps

AWS / Azure

Intermediate

Docker / Git

Advanced

Django / FastAPI

Expert

Work Experience

A timeline of my professional journey, highlighting key roles and accomplishments in the AI/ML field.

AI Engineer

Solyntra

Dec 2025 - Present

  • Developed RAG-based enterprise wiki system with source attribution and hallucination prevention, ensuring trustworthy knowledge retrieval
  • Implemented "no source, no result" verification pipeline for factual grounding in organizational knowledge bases

ML Engineer

Programming Hero

Dec 2023 - Sept 2025

  • Zenyora AI Wellness: Owned full ML lifecycle for production platform serving 100+ daily users with 95% posture detection accuracy, reducing eye strain by 40%
  • Real Estate Sales Agent: Built voice-based RAG agent handling 500+ daily inquiries with 95% query relevance, increasing lead engagement 3x
  • Student Performance Prediction: Forecasting system serving 10K+ students with 85% accuracy, contributing to 15% grade improvement

AI Engineer

Primacy Infotech Ltd

July 2023 - Nov 2023

Led AI tour planner for 6 UNESCO heritage sites, reducing itinerary creation from 2 hours to 5 minutes. Deployed FastAPI + Streamlit system achieving 99.2% uptime, increasing tourist satisfaction by 35%.

Research Assistant

Rising Research Lab

Aug 2022 - May 2023

Developed multi-disease prediction system (diabetes, Parkinson's, maternal health) with 92% average accuracy. Engineered IoT air quality monitoring network across 5 locations with real-time inference.

AI Trainer

Enhancing Digital Government Economy (EDGE) Project

May 2023 - Nov 2023

  • Trained 50+ students in ML/AI through hands-on projects, developing a curriculum covering supervised learning, clustering, regression techniques, and model evaluation metrics
  • Created educational materials with clear explanations of complex ML concepts, demonstrating strong communication skills
  • Evaluated student ML models, providing detailed feedback on error modes and improvement strategies

Production Projects

A selection of projects that demonstrate my ability to build and deploy robust, end-to-end machine learning solutions.

OgroPath Medical Entrance Exam Platform

OgroPath - Medical Exam Prep SaaS

FOUNDER

AI-powered medical entrance exam preparation platform serving students across Bangladesh. Features NLP-based question generation, 21K+ question bank, adaptive study plans, and AI diagnostic engine analyzing performance to generate personalized assessments.

LLMs NLP FastAPI React
M-TRUST Medical Fairness Toolkit

M-TRUST: Medical AI Fairness Toolkit

Open-source Python package for detecting and mitigating bias in medical AI. Published on PyPI with 1K+ downloads and 30.8% bias reduction.

Python scikit-learn PyPI
ExplainRAG Clinical Decision Support

ExplainRAG-FC-AS: Clinical Decision Support

RAG-based system with thematic clustering and NLI fact-checking. Achieves <10s response time with 90% clinician preference.

LangChain FAISS GPT-4
Zenyora AI Wellness Platform

Zenyora AI Wellness Platform

LIVE

Production wellness platform serving 100+ daily users. Real-time posture detection using MediaPipe (95% accuracy, 40% eye strain reduction), context-aware productivity coaching with sub-100ms latency processing 30 FPS video.

PyTorch MediaPipe GPT-4 FastAPI
web

RAG-Based Website Generator

Cost-efficient AI website generator achieving 94% cost savings ($0.03 vs $0.50-$2.00) and 92% token reduction vs v0/Lovable. Uses RAG with PostgreSQL + pgvector for semantic search across 21 indexed React components, tree-sitter for AST parsing.

RAG PostgreSQL GPT-4o Next.js
translate

Bengali Literature RAG System

RAG pipeline combining LaBSE embeddings with BM25 keyword search achieving 82% MRR, 95% Hit@5 accuracy with 32ms response time. Intelligent GPT-4o Mini enhancement reducing LLM API costs by 60% while maintaining quality.

LaBSE BM25 GPT-4o Mini FastAPI

Professional Milestones

Key achievements and recognitions throughout my career

Industry Networking

Industry Networking

Professional visit to Bangladesh Association of Software and Information Services

Corporate Recognition

Corporate Recognition

Recognition from Robi Axiata for exceptional performance and problem-solving skills

AI Competition

AI Competition

Participating in AI Olympiad, demonstrating cutting-edge ML expertise