AI Developer with 5+ years of experience building production ML and Data pipelines, computer vision models, and NLP systems — from EU Horizon research to deployed products.
I'm Fahim Zafri — an AI Developer based in Bangladesh. I hold a BSc in Civil Engineering from BUET and have spent the last 5+ years applying machine learning to real-world problems: flood prediction, NLP pipelines in Environmental Resilience, computer vision, and agentic AI systems.
Currently I'm an AI Engineer at Taskatask where I am building a next generation AI platform for task and workflow management, I have also worked at Technovative Solutions Ltd. as a Junior Researcher Intern for 8 months where I contributed to the EU Horizon projects (CLIMAS, C2IMPRESS). I've worked across all product stages — from Ideation, Product Design to Data Engineering and System Architecture that can scale.
I also teach. As an Specialist AI Trainer at AI for All (Send a Little Hope Foundation), I run courses on Python, Data Science, and FastAPI for both STEM and non-STEM audiences — because good AI literacy matters beyond engineering teams.
A selection of work spanning research, tooling, and production systems.
U-Net CNN trained on Sentinel-1 SAR imagery, geospatial, and rainfall data to predict flood extents in near real time. Achieved 97% accuracy, 92% F1-score, and 93% AUC at 10 m resolution.
NLP and transformer-based system to generate Climate Assembly agendas from academic, social, and news data. Integrated a formal ontology and SWRL/SPARQL symbolic reasoning engine to reduce model hallucinations.
LangChain-based chatbot with fine-tuned LLMs, prompt engineering, and RAG for personalised learning. Scraped and curated training data; optimised for latency and answer quality.
Semantic search engine for Bangladesh clothing and wearables stores. Uses NLP embeddings to match natural language queries to products across multiple local retailers.
Apache Airflow pipeline that extracts product prices with Selenium, processes and stores data in AWS S3, and enables real-time price tracking across multiple e-commerce platforms.
Exploratory data analysis on Dhaka Stock Exchange company prices. Applies statistical methods to identify optimal buy/sell windows from opening and closing price data.
Try a vision model running live in your browser, or send text to a model served by the backend.
Open to freelance ML/AI projects, collaborations, and consulting engagements. Reach me at zafri.fahim@gmail.com.
Book a free 30-min call →