Md. Asif Iqbal Fahim

College Road · Sreemangal · Bangladesh · (+880) 1303-021134 · asif.iqbal.fahim.bd@gmail.com
Software Engineer (Python/AI) | Machine Learning Engineer | Deep Learning Specialist | GenAI Expert

  • Seasoned Machine Learning Engineer with over 5 years of comprehensive experience in Machine Learning, Deep Learning, Generative AI, and Large Language Models. Proven track record across diverse environments including international remote positions, research institutes, and leading software companies.

  • Graduated from the Green University of Bangladesh with a major in Computer Science & Engineering in 2020.

  • Had Research Publications in high impact-factor Journal, Conference on Computer Vision domain.

  • Attained Kaggle Expert(X2) Title(Top 0.8%). Achieved 3 Silver Medals & 1 Bronze Medal. Selected as DataScience Advisor for KaggleX BIPOC Program 2024.

  • Selected as a Mentor & Alpha tester by Deeplearning.ai for Natural Language Processing in Coursera Platform.

  • Achieved Regional Champion & Global Nominee Title in NASA Space Apps Challenge 2020. Team SiliconLily has received the following awards and nominations: Global Nominee.

  • Participated in many Competitive Coding Contests (Regional ACM ICPC, NCPC, Inter-University Programming Contests, etc.)







Areas of Expertise

  • Generative AI & LLMs: RAG Systems, Prompt Engineering, Model Fine-tuning (PEFT, LoRA, QLoRA), LangChain, Vector Databases

  • Machine Learning & Deep Learning: TensorFlow, PyTorch, JAX, ONNX optimization, Model deployment and scaling

  • Speech & NLP: STT/TTS systems (Whisper, wav2vec2), NLP pipelines, Transformer models, BERTopic

  • MLOps & Infrastructure: Docker, Kubernetes, CI/CD pipelines, AWS, Terraform, Microservices architecture

  • Programming: Python (Expert), C/C++, Java, Kotlin, JavaScript

  • Frameworks & Tools: Flask, FastAPI, Celery, Redis, Elasticsearch, OpenSearch



Research Interests

  • Machine Learning

  • Deep Learning

  • Computer Vision

  • Sequence Modeling

  • Natural Language Processing

Educations

B.Sc. in Computer Science & Engineering

Green University of Bangladesh , Bangladesh
CGPA: 3.566 out of 4.00 | Ranked 2nd in Batch.
Two times received Dean Awards for Outstanding Academic Performances.

Higher Secondary Certificate (H.S.C.)

Sreemangal Govt. College, Sreemangal , Bangladesh
GPA: 5.00 (out of 5.00).

Secondary Secondary Certificate (S.S.C.)

The Buds Residential Model School, Sreemangal , Bangladesh
GPA: 5.00 (out of 5.00).

Certifications

Deep Learning Specialization - Deeplearning.AI

Platform : Coursera - Credential
May. 2020

AWS Machine Learining Nanodegree Program

Platform : Udacity - Credential
May. 2020

AI for Medicine Specialization - Deeplearning.AI

Platform : Coursera - Credential
Jan. 2021

Natural Language Processing Specialization - Deeplearning.AI

Platform : Coursera - Credential
Jan. 2021

Generative Adversarial Networks (GANs) Specialization - Deeplearning.AI

Platform : Coursera - Credential
Feb. 2021

TensorFlow Developer Specialization - Deeplearning.AI

Platform : Coursera - Credential
August. 2020

Android App Development Specialization - Vanderbilt University

Platform : Coursera - Credential
August. 2020

Device-based Models with TensorFlow Lite - Deeplearning.AI

Platform : Coursera - Credential
June. 2020

Basic Image Classification with TensorFlow - Coursera Project Network

Platform : Coursera - Credential
May. 2020

Custom object detection on Google colab & android deployment

Platform : Udemy - Credential
Dec. 2019

Intro to ML: Image Processing

Platform : Qwiklabs - Credential
Jan. 2020

Intro to ML: Language Processing

Platform : Qwiklabs Credential
Jan. 2020

Computer Vision - Object Tracking with OpenCV and Python - Coursera Project Network

Platform : Coursera - Credential
Apr. 2020

Competitive Programming

Platform : Coding Blocks Credential
Feb. 2020

Experience

Software Engineer (Python/AI) [Full Time - Remote]

InfinitiBit, Germany

Working as Software Engineer focusing on Python/AI projects.

  • Led the development of an innovative Multimodal Agentic RAG system, implementing advanced PDF processing capabilities for extracting and analyzing text, tables, images, and diagrams using unstructured.io and ChromaDB
  • Collaborated with Siemens on developing intelligent energy solutions, creating AI Agent Workflows for grid optimization and predictive maintenance applications
  • Orchestrated AWS infrastructure using Terraform for automated EC2 instance scaling and management, optimizing resource allocation for AI agent workflows and ensuring high availability
  • Developed and documented a metadata-aware RAG system with embedding caching strategies using Redis and semantic caching to optimize OpenAI API usage and reduce costs
  • Implemented causal reasoning models and inference frameworks for time-series analysis, enabling more robust prediction capabilities in renewable energy forecasting applications
  • Designed and deployed AI agent workflows for automated data processing pipelines, incorporating customized reasoning engines and decision-making frameworks
  • Spearheaded optimization efforts for PyPSA (Python for Power System Analysis) library, including code profiling, debugging, and performance enhancement
  • Implemented proof-of-concept for LLaMA 7B (PEFT) embedding extraction, enhancing the system's natural language processing capabilities

December 2024 - Present

Deep Learning & GenAI Instructor [Part Time]

aiQuest Intelligence, Bangladesh

Providing advanced training in Generative AI and Deep Learning.

  • Led advanced training programs in Generative AI, focusing on LLMs, multimodal AI systems, and state-of-the-art model optimization techniques (PEFT, LoRA, QLoRA)
  • Instructed practical implementation of modern AI architectures including RAG systems, RLHF, and LLM preference alignment with hands-on projects
  • Mentored students in developing end-to-end projects including YOLO-based traffic detection systems and LLM-powered chatbots using Ollama, LangChain, and Vector Databases
  • Provided guidance on industry-standard practices in model optimization, including quantization techniques, knowledge distillation, and parameter-efficient fine-tuning

September 2024 - Present

Software Engineer - Machine Learning/Recommendation Systems [Full Time]

IdeaScale Bangladesh Limited

Worked as Software Engineer focusing on ML and Recommendation Systems.

  • Developed end-to-end recommendation systems using ALS-based matrix factorization and implemented Context-Aware Recommendation Engine with Elasticsearch for personalized user experiences
  • Optimized LLM deployments using PEFT, LoRA, and ONNX-accelerated Sentence-BERT, while implementing OpenSearch Vector Database with HNSW for efficient semantic search
  • Implemented adaptive BERTopic modeling with incremental training using IncrementalPCA and MiniBatchKMeans, deployed via Celery-powered parallel processing pipelines
  • Architected and deployed ML solutions as microservices using Docker, Kubernetes, and Flask RESTful APIs, while mentoring junior team members

December 2023 - November 2024

Machine Learning Engineer [Full Time]

REVE System Ltd

Worked as Machine Learning Engineer focusing on Speech AI.

  • Developed and optimized Speech-to-Text (STT) solutions using OpenAI Whisper & wav2vec2, creating phoneme and grapheme models with dataset-specific vocabulary enhancement
  • Implemented various Text-to-Speech (TTS) architectures (Align TTS, GlowTTS, FastSpeech2, vits, ProDiff) with multiple vocoders (HiFiGAN, MultiBandMelGAN, WaveRNN, UnivNet)
  • Enhanced model performance through Speech Augmentation techniques and dynamic training, leveraging JAX framework, Peft, and LoRA for large-scale optimization
  • Developed WhisperX streaming service with voice activity detection and forced phoneme alignment, deploying containerized solutions via FastAPI and Flask

December 2022 - December 2023

Research & Teaching Assistant [Full Time]

RIoT Center Research LAB & Independent University of Bangladesh

Worked as Research & Teaching Assistant in this Institution.

  • Worked on Research project: Response of Bevacizumab effectiveness in Ovarian Cancer from WSI Dataset w/ fine-tuned State-of-the-art ViT model & achieved 96% Test Accuracy & 99% ROC AUC Score. (Accepted in ICACIS 2024 and publication to the Springer Nature Singapore (Scopus indexed CCIS Series))
  • Taught the course under the CSE department of the Independent University of Bangladesh, - CIS 101: Fundamentals of Computer Systems LAB for three sections total students 40/Class.

September 2022 - 14th December 2022

Research Assistant [Full Time]

Institute of Advanced Research (IAR) LAB, United International University

Worked as Research Assistant in this Institution.

  • Worked in Research project, "Semantic Biomedical Image Segmentation Using Attention Mechanism Driven Modified U-Net Based on Deep CNN" (funded by the Institute of Advanced Research, UIU, Grant Reference: UIU/IAR/02/2021/SE/25.) (Published in High Impact Factor Journal(6.0) Neurocomputing, Elsevier)

February 2022 - August 2022

NLP Specialization Mentor & Alpha Tester

Deeplearning.AI, USA

Responsible Mentor & Alpha Tester for Natural Language Processing Specialization in Coursera Platform. Achieved Outstanding Mentor Title.

  • Tested curriculum before published to the Coursera & Provided feedbacks to course staffs. Guide learners if didn't understand the course contents.

February '21 - May '21

Junior Research Assistant [Part Time]

Consortium Research Center (Previously ResearchHub)

Responsible Researcher in Machine Learning Domain.

  • Did Research on Efficiently Convert Time Series Data to Image & Deep Learning Modeling for Credit Card Fraud Detection.

Sep 2021 - Mar 2022

Data Scientist - Independent Contractor

Dentalfynd (Previously ilekun Health), USA

Worked as a Data Scientist in this StartUp company.

  • Aggregate NYC borough's dentist data by web scraping (Selenium & Scrapy), Data cleaning, Predictive modeling.

May 2021 - July 2021

Data Science - Trainee

CRIDA, Canada Based Organization

Trained in advanced Deep Learning techniques.

  • Build Attention Based LSTM Networks, Used TPU Strategy - Distributed Training for a large amount of Dataset, Used Frameworks Like PyTorch, TensorFlow.

March 2021 - June 2021

Data Science & Business Analytics - Intern

The Spark Foundation

Worked in the Business Problem as a Data Science & Business Analytics Intern.

  • Performed Exploratory Data Analysis & Predictive Modeling with Machine Learining Algorithms.

Oct 2021 - Nov 2021

Machine Learning Developer - Intern

TechnoColabs

Worked in the Business Problem as a Data Science & Business Analytics Intern.

  • EDA ,Feature Engineering ,Missing Value Handlin for Spotify Track Skips.
  • Build Predictive Modeling for Predict Spotify Track Skips.
  • End to End Deployment for Spotify Track Skips.

Oct 2021 - Nov 2021

ML Achievements & Activities

Kaggle - The world's largest data science community

  • Selected as Datascience Advisor for KaggleX BIPOC Program, 2024, Goal: Help fellows to fine-tune Google LLM Gemma Model for downstream task.
  • Selected as Datascience Mentor for KaggleX BIPOC Program, 2023, Goal: Help mentee to complete end-to-end data science projects.
  • "Expert" in Kaggle ML/DL Competitions - Rank Top 0.8% among 182K+ & "Expert" in Kaggle Discussions Forum - Rank Top 0.3% among 269K+ active Kaggle users worldwide.
  • Bengali.AI Speech Recognition Comp., Kaggle - Silver Medal Rank in Top 6%
  • PetFinder.my - Pawpularity Comp., Kaggle - Silver Medal Rank in Top 2%
  • G-Research Crypto Forecasting Comp., Kaggle - Silver Medal Rank in Top 5%
  • G2Net Gravitational Wave Detection Comp., Kaggle - Bronze Medal Rank in Top 10%
  • Bristol-Myers Squibb – Molecular Translation Comp., Kaggle - Rank in Top 13%
  • Cassava Leaf Disease Classification Comp., Kaggle - Top 16%.
  • Dhaka AI Traffic Challenge Comp., GUB- Finished at 45th.

Sep 2020 - Present

Data Science Stack Exchange Contributor

Data Science Stack Exchange - The world's best Q&A community

[Data Science Stack Exchange Profile]

  • Top 6% Contributor to Data Science Stack Exchange in 2021. 13th Q. Answered, 1 Gold, 3 Silver, 14 Bronze Medals achieved.
  • AWS Machine Learning Scholarship Program, Udacity - Scholar

Feb 2021 - Present

Android Application Developer - Intern

Leotech Ltd.(Govt. Project)

Developed various Android Applications using Native Android Development. [Feature Project]

  • Worked with MVVM Pattern, Rest API, Firebase, MLKit, Google Sceneform in fast-paced environment, wrote maintainable and reusable Java and Kotlin Code.

November 2019 – May 2020

Core Team Member

International AI-based Dhaka Traffic Detection Challenge

Core Team Member in the International Event. [Core Team Member Github Organization Page]

  • Responsible for acquisition of the top solutions from the Challenge.
  • As contribution integrate Flask JSON Web Token in Website for Security.

Oct 2020 - Dec 2020

Projects


  • Camera Beautify - AI Face Recognized Android Application

    This is the bangladeshi traditional AI based Camera Stickers and Face Beautify Android Application. To build this application maintained agile methodology and software engineering best practices. ( Project Link)
    Architectural Pattern & Language Used : MVC & Java.
    Technologies used : Open CV , Open Graphics Library (OpenGL®) , HTML ,and Many Android Components like recyclerview , webview , shared preference etc.

  • New Connection Chat Bot

    This is basically the Telegram Chat Bot. This conversational chat bot help depressed peoples. Google Dialogflow integrated into the software to became more intelligent. ( Project Link)
    Architectural Pattern & Language Used : MVC & Python.
    Hosted & Services Used : Heroku & Telegram API.

  • Multi Class Image Classification Using SVM

    This was performed on Kaggle's dogs-cats-horses-humans Dataset, using Support Vector Machine model for Classifying Dog or Cat or Human or Horse. ( Project Link)
    Language used: Python.
    Development Tools: Jupyter Notebook.

  • Predict Profit for Bike Sharing Company

    This was performed on a Tabular dataset which contains Profit and Population attributes. Linear Regression model used for predict future profit of this company. ( Project Link)
    Language used: Python.
    Development Tools: Jupyter Notebook.

  • Customer Car Purcheased Group Classification and Prediction

    This was performed on a Tabular dataset which contains Gender,Age and Estimated Salary attributes. Support Vector Machine model used for Classify and Predicts various of Car Purchased Group. This project involves data pre-processing, Feature Scaling, and Statistical Analysis Like confusion metrics. ( Project Link)
    Language used: Python.
    Development Tools: Jupyter Notebook.

  • My71 - Android application about Bangladesh Liberation War

    This app encourage peoples by sharing about Bangladesh liberation war's histories. Maintain user-friendly features. ( Project Link)
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : Expandable List View , Foldable Layout , Gif ImageView , Recycler View and ohter android conponets.
    Development Tools: Android Studio.

  • Cat and Dog Recognition - Android application for recognizing cats or dogs

    This is the cat and dog detection android application. This is a camera app that continuously detects the objects (bounding boxes and classes) in the frames seen by device's back camera, using a quantized MobileNet SSD model trained on the COCO dataset. ( Project Link)
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : TensorFlow Lite and Many Android Components like recyclerview , webview , shared preference etc.
    Development Tools: Android Studio.

  • Space Air Quality Monitoring - Nasa Space App Challenge 2019 Project - (Android Version)

    This applicaton shown the details of the air quality of the surface from IOT Device Data through API. And analysis those data with NASA Data. ( Project Link)
    See the Details Description From : NASA
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : Rest Api , Expandable List View , Foldable Layout , Recycler View and ohter android conponets.
    Development Tools: Android Studio.

  • Hello AR - Augmented Reality related Android Application

    This is fun app . Using this app user can set augmented fox whatever user want. ( Project Link)
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : Google Sceneform , Android Components like recyclerview , shared preference etc.
    Development Tools: Android Studio.

  • Corona Informative Android App

    This applications provide information about Covid 19 for Bangladesh. ( Project Link)
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : Expandable List View , Foldable Layout , Gif ImageView , Recycler View and ohter android conponets.
    Development Tools: Android Studio.

  • News Hub - A Online News App

    By using this application users get all online newspaper in one app. It's user friendly application. Trying to satisfy the users need. ( Project Link)
    Architectural Pattern & Language used: MVC & Java
    Technologies Used : Android Components like Recyclerview , Webview , Shared preference etc.
    Development Tools: Android Studio.

  • Environment Natural Events - Nasa Space App Challenge 2020 Project (Champion Project)

    This project achived Global Nominee from Nasa Space Apps Challenge. This website is great visulizaion project for nearby natural events. ( Project Link)
    See the Details Description From : NASA
    Architectural Pattern: MVC
    Technology & Language used : Node.js , React js , Leaflet js , Chart js , Axios

  • F & E Shoppy - E-Commerce Website

    This is the E-Commerce Website with Admin Panel and User Panel. ( Project Link)
    Architectural Pattern: MVC
    Language used : HTML , CSS , bootstrap , PHP and MYSQL

  • Space Air Quality Monitoring - Nasa Space App Challenge 2019 Project - (Web Version)

    Through This is the website, We shown the details of the air quality of the surface from IOT Device Data. And analysis those data with NASA Data. ( Project Link)
    See the Details Description From : NASA
    Architectural Pattern: MVC
    Language used : HTML , CSS , bootstrap , Java Script and MYSQL

  • Snack Game - JS Edition

    This is a fun type snack game. Where snack always eats the apples and want to go big. ( Project Link)
    Architectural Pattern: MVC
    Language used : Javascript & HTML
    Technologies used : Javascript Phaser

  • Space Air Quality Monitoring - Nasa Space App 2019 IOT Source Code

    This code used for detect harmful gases from environment and send them to the website and android using web server. ( Project Link)
    See the Details Description From : NASA
    Language used: C/C++
    Hardware Used : NodeMCU and Sensors
    Development Tools: NodeMCUSimulator and Arduino IDE.

  • Temperature Indicator for baby feeder

    By using a waterproof temperature sensor to get the reading of the liquid’s temperature. Based on that temperature taking decision whether the temperature is too hot for the baby to drink or too cold. ( Project Link)
    Project platform: Arduino
    Equipment: Arduino Uno R3 , DS18B20 temperature sensor , Colour (RGB)LED , Jumper wires, Resistors ( 220ohm, 1k ), Buzzer
    Development Tools: NodeMCUSimulator and Arduino IDE.
    Programming Language: C/C++

  • Multimodal RAG System with Intelligent Caching - InfinitiBit

    • Architected production RAG system processing PDFs with text, tables, images using ChromaDB vector storage
    • Implemented Redis-based semantic caching reducing OpenAI API costs by 60%, deployed with Terraform on AWS

  • Real-time Speech Recognition Streaming Service - REVE Systems

    • Developed WhisperX streaming service with WebSocket support, voice activity detection, and phoneme alignment
    • Built containerized deployment with FastAPI, achieving sub-200ms latency for real-time transcription

  • Scalable Recommendation Engine with Vector Search - IdeaScale

    • Built context-aware recommendation system using OpenSearch vector database with HNSW indexing
    • Deployed ONNX-optimized Sentence-BERT models achieving 5x latency reduction in production

  • Distributed ML Training Infrastructure - REVE Systems

    • Designed dynamic training system for 900-hour speech dataset with automated GPU scheduling
    • Implemented JAX-based optimization pipeline reducing training time by 40% using model pruning

  • Incremental Learning Pipeline with BERTopic - IdeaScale

    • Built production pipeline for topic modeling with incremental training using IncrementalPCA and MiniBatchKMeans
    • Implemented Celery-based distributed processing handling 10K+ requests/minute with automated scaling

  • AI Energy Grid Optimization - InfinitiBit/Siemens Collaboration

    • Developed AI agent workflows for predictive maintenance and grid optimization
    • Implemented time-series analysis with causal reasoning models for renewable energy forecasting

  • CI/CD Pipeline for ML Model Deployment

    • Built end-to-end CI/CD pipeline using GitHub Actions for automated model testing, validation, and versioning
    • Integrated AWS CodePipeline for automated deployments with rollback capabilities and A/B testing support

  • Bengali.AI Speech Recognition - Kaggle Competition

    • Silver Medal (Top 6%): Developed optimized speech recognition models with production-ready data pipelines

  • AI Face Recognition Android App - Edge Deployment

    • Deployed MobileNet SSD on mobile devices using TFLite with real-time OpenCV inference

Research Publications & Activities

Research Activities

  • Active Reviewer for Elsevier Journals (October 2024 to May 2025): Completed 11 reviews for high-impact journals - Neurocomputing (5 reviews), Natural Language Processing Journal (3 reviews), and Engineering Applications of Artificial Intelligence (3 reviews).
  • Selected Reviewer for a Renowned International Conference 6th Intl. Conf. on Imaging, Vision and Pattern Recognition (IVPR) and 11th Intl. Conf on Informatics, Electronics & Vision (ICIEV).

Publications Gallery

Improved Transfer Learning Architecture to Classify Covid-19 Affected Chest X-Rays using Noisy Student Pre-training
Md. Asif Iqbal Fahim, Feroza Naznin, Mohammad Ali Moni, and Dr. Md Zahidul Islam
ICIEV & ICIVPR 2021, JAPAN
Access Publication
COVID-19 Research
DOLG-Next: Deep Orthogonal Fusion of Local and Global Features with ConvNext for Biomedical Image Segmentation
Md. Rayhan Ahmed, Md. Asif Iqbal Fahim, A.K.M. Muzahidul Islam, Salekul Islam, Swakkhar Shatabda
Neurocomputing, Elsevier (IF: 6.0)
Access Publication
DOLG-Next Research
Analysis of AI models to Determine the Effect of Bevacizumab Therapy on Epithelial Ovarian Cancer from Histopathology Images
Md Shakhawat Hossain, Md. Asif Iqbal Fahim, Munim Ahmed, Md Sahilur Rahman, Joynob Ema, Rubina Bari, Mahbubul Syeed, Razib Hayat Khan, Faisal M Uddin
ICACIS Conference 2024
Access Publication
Ovarian Cancer Research

Skills

Programming Languages
Machine Learning & AI Frameworks
Web Development & Tools
Cloud & DevOps
Databases & Big Data
Technical Skills

Workflow & Methodologies

Awards & Achievements

Administrative Experience

Voluntary Experience