Hello, I am

Farhin Rahman

AI Engineer • Researcher
Download CV

About me

AI Engineer • CS Graduate

AI researcher and engineer with publications in healthcare AI and NLP. I design and build end-to-end intelligent systems—from agentic workflows with LangGraph/LangChain and fine-tuned transformers to robust data infrastructure with MLOps best practices.

Personal Info

  • Occupation : AI Engineer
  • Email : sadiafarhin62@gmail.com
  • Phone : +880 1751 071851
  • University : BRAC University
  • Address : South Banasree, Dhaka-1219

Areas of Expertise

Agentic AI Systems

LangGraph & LangChain for multi-step orchestration, RAG pipelines (vector + graph‑augmented), tool-calling, and multimodal agents (text+image+PDF).
Stack: LangGraph, LangChain, LlamaIndex, Qdrant, Neo4j, FAISS, SQL AST validation.

Machine Learning & MLOps

Deep learning for NLP/CV; deployment to production with robust pipelines.
Stack: PyTorch, HuggingFace Transformers, Scikit-learn, MLflow, Weights & Biases, Docker, CI/CD (GitHub Actions), data versioning (DVC). Databases: PostgreSQL, Vector DBs (Qdrant, Pinecone, Weaviate ), Graph DB (Neo4j).

Software & Cloud Engineering

Full‑stack systems with scalable APIs, frontend frameworks, and cloud‑native infra.
Stack: Python (FastAPI, Flask), Node/Express, JavaScript/TypeScript, React, Next.js, SQLAlchemy, REST/GraphQL APIs. Cloud: AWS (EC2, S3, Lambda), Docker, Kubernetes.

My Resume

Work Experience

AI Engineer • Kodamai

Aug 2025 – Present

Agentic System Design & Orchestration

  • Prototyped a conversational agent MVP using LangGraph/LangChain with robust tool-calling and evaluation loops.
  • Co-designed multimodal data architecture: Vector (Qdrant), Graph (Neo4j), and SQL for complex retrieval.
  • Built ingestion pipelines leveraging embeddings and graph relationships for entity matching.
  • Established CI/CD, containerization, tracing/profiling for scalable performance.
AI Engineer • Ontik Technology

May 2025 – Jul 2025

Social Media Intelligence (Banking)

  • Automated sentiment analysis pipeline on public Facebook data via Apify ingestion.
  • Fine-tuned multi-class sentiment model; added priority routing for critical items and dashboard reporting.
Software Engineer (AI/ML) • micro1

Mar 2024 – Apr 2025

  • Fine-tuned AI interviewer; integrated spaCy NER & SBERT scoring; cut processing time by 40%.
  • Optimized pipelines via caching/batching (−50% data prep, +30% training efficiency).
  • Set up MLflow MLOps lifecycle; reduced deployment time by 60% and production bugs by 45%.
Junior Software Developer • BCIC Swiss GmbH

May 2023 – Sep 2023

  • Developed full-stack web apps (React/JS) and REST APIs (PHP) in production.
Research Assistant (Part-time) • BRAC University

Aug 2022 – Dec 2023

  • Contributed to NLP/CV research: datasets, experiments, and co-authored manuscripts.

Education

BRAC University • B.Sc. in Computer Science

2018 – 2022

Thesis: Predicting Brain Age from EEG Signals using ML & Neural Networks

  • Honors: VC’s List (Top 2% — Summer 2020)
  • Dean’s List (Top 5% — Spring 2019)
Motijheel Ideal School & College • HSC (Science)

Obtained Golden GPA 5.0; awarded scholarship in the talent pool.

Motijheel Model High School & College • SSC (Science)

Obtained GPA 5.0 with an average of 94% marks.

Skills

Python
JavaScript / TypeScript
SQL
Docker & Linux
PyTorch / Transformers
MLflow & MLOps

Languages

English
Bengali
French
Japanese

Hardworking

Motivated

Enthusiastic

Tech-savvy

My Portfolio

All Projects Certificates Publications
Multilingual AI Support Agent

AI financial assistant with SQL AST validation, multi-RAG retrieval (qualitative/logical/numeric), and neuro-symbolic rules for ranked recommendations, enhancing decision-making and user interactions.

Stack: Python, LangGraph, Streamlit, Ollama, SQLite, yfinance/Stooq, RAG
Live app
LinkToBD — Social & Networking Platform

Versatile social platform connecting people across networking, business, travel & more. Built during my tenure at BCIC, it features dynamic content and user-driven interaction for seamless connections.

Stack: React, PHP, JavaScript, REST API
Visit Site
Refili — Recyclable Shopping Bags

A fully responsive e-commerce platform for recyclable jute bags, built with a React.js frontend for interactivity and PHP, JavaScript backend for server-side operations.

Stack: React.js, PHP, JavaScript, Bootstrap
Visit Site
H-Food

A sentiment analysis model for Bangladeshi food-delivery reviews, leveraging PyTorch, Transformers to handle class imbalance. Sklearn was used for preprocessing and feature engineering.

Stack: Python, PyTorch, Transformers, Sklearn
Visit GitHub
Grape Leaf Diseases (XAI)

A CNN model detecting grape leaf diseases with a 99.7% F1-score. Keras powered the model, with LIME for explainability to highlight key features. It improves precision in early disease detection.

Stack: Python, Keras/TensorFlow, LIME, OpenCV
Visit GitHub
Banking Sentiment Analysis

Automated sentiment analysis pipeline for a major bank, using Apify to gather social media data and a multi-class sentiment model to classify public feedback (positive, negative, inquiry).

Stack: Python, OpenAI API, Streamlit, Apify
Visit App
ICCTA 2023 Certificate
ICCTA 2023 Presenter

Presented paper on "Early Autism Disorder Detection Through Visualizing Eye-Tracking Patterns Using CCT" at the International Conference on Computer Technology Applications in Vienna, Austria.

Event: ICCTA 2023, Vienna, Austria
View Paper
IEEE IICAIET 2022 Certificate
IEEE IICAIET 2022 Presenter

Presented paper titled "Bengali Speech Recognition: An Overview" at the 4th IEEE International Conference on Artificial Intelligence in Engineering and Technology, held virtually.

Event: IEEE IICAIET 2022, Virtual Conference
View Paper
Full Stack Development Certificate
Full-Stack Certification

Certified by micro1 for excellent knowledge of Full Stack Development, covering modern web technologies including React, Node.js, databases, and API development.

Issuer: micro1
Date: November 18, 2024
Data Science Certificate
Data Science Certification

Certified by micro1 for excellent knowledge of Data Science, including machine learning, statistical analysis, data visualization, and predictive modeling.

Issuer: micro1
Date: February 27, 2024
Early Autism Disorder Detection (CvT)

Eye-tracking pattern visualization using Compact Conv. Transformers for ASD support.

Read paper
Transfer Learning for Waste Classification

VGG16, MobileNetV2, and a 6-layer CNN baseline on a waste dataset.

Read paper
Bengali Speech Recognition: An Overview

Survey of datasets, methods, and challenges for Bengali ASR.

Read paper