About Experience Research Projects Contact

AI DEVELOPER & RESEARCH ASSISTANT

Abir Ehsan
Evan

Computer Science Graduate passionate about Machine Learning, Data Analysis, and building LLM-powered solutions for industrial and research applications.

Abir Ehsan Evan
0

Years Experience

0

Publications

0

Certifications

0

Projects

About Me

I'm a Computer Science graduate from Independent University Bangladesh with a deep passion for Machine Learning, Data Analysis, and Software Development. My work focuses on building intelligent systems that bridge the gap between cutting-edge AI research and practical industrial applications.

Currently, I'm working on enterprise AI systems including multimodal RAG chatbots, automated HR orchestration platforms, and voice-enabled AI assistants. My research experience spans from ethnographic HCI studies to astrophysics data analysis.

Education

BSc in Computer Science

Independent University Bangladesh

2020 - 2025

A' Levels - Physics, Mathematics

Academia

2019

Work Experience

AI Developer

Betopia Group

January 2026 - Present
  • Developed enterprise AI systems including a multimodal RAG chatbot for customer support, Automated HR orchestration and a hotel IVR concierge.
  • Engineered industrial-grade software solutions with OpenAI function calling and FAISS vector indexing, integrating AI assistants with SQLite databases.

Research Assistant

DND Lab, University of Dhaka

October 2025 - Present
  • Implemented multilingual voice-enabled system by integrating STT/TTS capabilities with ASR models.
  • Conducted ethnographic research at 3 zonal electricity offices, with HCI-oriented insights informing the agent's design.

Machine Learning Intern

Innovative Skills LTD, Bangladesh

May 2025 - August 2025
  • Developed a medical chatbot Web Application using NLP and AI models (OpenAI GPT / LLaMA) for symptom analysis.
  • Built a Django REST backend to integrate AI insights with scraped data for real-time symptom checking.

Research & Publications

Current Research

Currently immersed in advanced research focusing on:

  • Large Language Models (LLM)
  • Ethnographic AI Studies
  • Speech-to-Text (STT) Systems
  • Text-to-Speech (TTS) Models

Selected Publications

01

A Quantized and Lightweight CNN for Multiclass Object Detection of Malaria Parasites

IEEE Xplore

View Paper →
02

KSiFoldNet: A K-Symplectic Integrated Transformer Network for RNA Structural Prediction

IEEE Xplore

View Paper →
03

FSG: A Conditional GAN for Censoring-Aware Survival Prediction

IEEE Xplore

View Paper →

Featured Projects

A selection of AI systems, full-stack products, automation workflows, and tools.

01 AI / ML Systems

NexusAI Backend

Multi-Agent

Multi-agent business intelligence platform combining invoice OCR, RAG document retrieval, and sentiment analysis. Built with FastAPI, Groq AI, PostgreSQL, and Qdrant vector DB.

FastAPI Groq AI Qdrant DistilBERT + LoRA
View on GitHub →

ResearchMind AI

RAG + Graph

Multi-agent RAG platform for document intelligence. Features LangGraph 4-agent pipeline, Neo4j knowledge graph with contradiction detection, JWT RBAC, and LLM-as-judge evaluation achieving perfect faithfulness scores.

LangGraph FAISS Neo4j Mistral-7B
View on GitHub →

BizPulse AI

Vision OCR

Invoice tracking API that extracts vendor, amount, date, and items from images using Groq Vision. Stores records in PostgreSQL with weekly expense summaries and OpenClaw skill integration.

FastAPI Groq Vision PostgreSQL
View on GitHub →

Medical Chatbot with LLMs

LangChain + AWS

End-to-end medical chatbot with LangChain, Pinecone, and Flask. Deployed on AWS for scalable AI-driven symptom analysis and medical guidance.

LangChain Pinecone Flask AWS
View on GitHub →

Multi-Person Kinematic Behavior AI

Enterprise

Behavioral analytics engine using MediaPipe Tasks API with 32+ kinematic and geometric features including 3D joint angles and distance-normalized velocity for classification.

MediaPipe Python Behavioral Analytics
View on GitHub →

BioLab Studio Pro

3D + AI

Interactive microbiology and biotechnology learning platform with full 3D molecular visualization (Three.js), AI co-scientist system, simulation engine, and 8 complete educational modules from beginner to research level.

Three.js Node.js / Express AI Experiment Designer

02 Full-Stack Products

D&D Lab Research Platform

Production

Centralized lab coordination platform replacing scattered chat threads. Features RBAC (Faculty/Manager/Researcher), project & task management, global search, file uploads, comments, reactions, and notifications. Dockerized monorepo.

Next.js FastAPI PostgreSQL Docker
View on GitHub →

HR Forge

AI + Integrations

AI-powered HR operations platform with two modules: Workforce Hub (employee records, letter generation, WhatsApp coordination, audit logs) and Hiring Portal (AI job posting generation, LinkedIn + Facebook publishing).

Flask React + Vite OpenAI LinkedIn / FB APIs
View on GitHub →

Chatopia RAG Chatbot

Voice + RAG

Sophisticated RAG-based chatbot with PDF and image extraction, meeting scheduling, and voice-enabled interactions for seamless user experience.

RAG Voice AI Python

Premiere Bank Chatbot

Web Crawler + RAG

Full pipeline for crawling the Premiere Bank website, cleaning and chunking content for a knowledge base, and a Django-backed chatbot frontend with RAG-based retrieval.

Web Crawler Django RAG Pipeline
View on GitHub →

03 Automation Workflows

Cross-Channel Trial Expiry Orchestration

Automated Make.com workflow monitoring trial end dates, assessing engagement, and executing multi-channel responses (email, SMS, Slack) based on user activity level.

Make.com Google Sheets Gmail
View on GitHub →

Inbound Lead Qualification & CRM Routing

Automated pipeline processing 40-60 daily leads from an ROI calculator. Enriches, scores (Hot/Warm/Disqualified), and routes to CRM with instant sales notifications — eliminating 1hr of daily manual triaging.

Make.com Google Sheets CRM Gmail
View on GitHub →

AI Competitor Mention Tracker

Automates competitive intelligence by pulling mentions from Reddit and Hacker News, then generating structured briefings via Claude AI. Built for Deskflow to replace manual weekly monitoring.

Flask Claude API Reddit + HN APIs
View on GitHub →

Webhook Event Inspector & Campaign Dashboard

Real-time dashboard receiving, validating, and visualising webhook events from Customer.io, Appcues, and LinkedIn. Includes campaign health monitoring with configurable alert thresholds and a live event simulator.

Flask SQLite Real-time SSE
View on GitHub →

Certifications & Courses

Machine Learning Specialization

DeepLearning.AI & Stanford

Mar 2025

Unsupervised Learning, Recommenders, RL

Coursera

Mar 2025

AI and ML Fundamentals

Grameenphone Ltd

Mar 2025

Basic Data Analysis on STATA

Bangladesh Research Society

Feb 2025

Advanced Learning Algorithms

Stanford University

Feb 2025

Online Safety

Grameenphone Academy

Feb 2025

Facebook Ads Campaigns

Shikhbe Shobai

Nov 2024

Deep Learning & LLM Workshop

Dhaka University - CSE

Oct 2024

Python with Django Training

Easy Coding

Oct 2024

Classification Model Project

Akaademy

Sep 2024

AI in Bangladesh Summit 2024

Certificate of Appreciation

Sep 2024

Supervised Machine Learning

DeepLearning.AI & Stanford

Sep 2024

Claude Code in Action

Anthropic

May 2026

Verify →

Introduction to Model Context Protocol

Anthropic

May 2026

Verify →

Claude with the Anthropic API

Anthropic

May 2026

Verify →

Introduction to Agent Skills

Anthropic

May 2026

Get In Touch

I'm currently open to new opportunities and collaborations. Whether you have a question, a project idea, or just want to connect, feel free to reach out.