Browse categories
Explore
Fiverr Pro
English
$
USD
I will build nlp, bert models, ai agents, and rag, llm solutions for you using python
Data Science and Artificial Intelligence
Hi, I'm Sharafat Ahmed, a CSE student specializing in Data Science & Artificial Intelligence at East West University. I help businesses create professional AI-generated product images and videos for m...
About this Gig
Looking for more than just a basic NLP script?
I build end-to-end intelligent text systems from classical NLP pipelines to fine-tuned BERT models and production-ready AI agents powered by LangGraph and LangChain. Whether you need a sentiment classifier, a domain-specific chatbot, or a full multi-agent LLM system, I deliver clean, documented, and deployable solutions.
What I Offer:
1. NLP & Text Analytics
Text Preprocessing tokenization, stopword removal, lemmatization (spaCy / NLTK)
Text Classification & Sentiment Analysis (Naive Bayes, SVM, Logistic Regression)
Named Entity Recognition (NER), Keyword & Keyphrase Extraction
TF-IDF, N-gram Analysis, Word Frequency, Co-occurrence Networks
Topic Modeling LDA, NMF, BERTopic
Text Summarization & Semantic Similarity
2. BERT & Transformer Fine-tuning
Fine-tune BERT, RoBERTa, DistilBERT, AraBERT on your custom dataset
Sequence classification, token classification, question answering
Training curves, evaluation report (accuracy, F1, confusion matrix)
Save & export model weights (HuggingFace format, .pth, .zip)
3. AI Agents & LLM Solutions
Multi-agent orchestration using LangGraph , Domain Specific
Programming language:
Python
•
MATLAB
•
SQL
•
Colab
Frameworks:
Scikit-learn
•
PyTorch
•
Panda
APIs:
Other
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
Excel
•
Colab
Other Data Science & ML Services I Offer
FAQ
Q1: What kind of text data can you work with?
Any domain — medical/clinical text, customer reviews, social media posts, YouTube comments, legal documents, academic papers, financial reports, survey responses. If you have text, I can build something with it.
Q2: Do I need a labeled dataset for classification?
For supervised tasks (classification, sentiment) — yes, labeled data is needed. For unsupervised tasks (topic modeling, clustering, keyword extraction) — raw text is fine. I can also advise on labeling strategy if you're starting from scratch.
Q3: Can you build a RAG system for my documents or knowledge base?
Yes — this falls under the Premium package. I'll set up a vector store (FAISS or Chroma), connect it to your documents, and build a LangChain retrieval pipeline so your LLM answers questions strictly from your data.
Q4: Which LLMs do you work with?
OpenAI GPT-3.5 / GPT-4, Groq (LLaMA 3, Mixtral), Google Gemini, Mistral. I can work with whichever you prefer or already have API access to. I can also use open-source local models via Ollama if you want zero API costs.
Q5: Will I be able to run and modify the code myself?
Absolutely. All deliverables are clean, well-commented Jupyter/Colab notebooks. I write code for humans, not just machines. You'll understand every step, and I'm happy to explain anything after delivery.
Q6: Can you deploy the model or agent as an API or web app?
Basic deployment (FastAPI endpoint or Streamlit app) can be added as an extra. For full cloud deployment (AWS, GCP, Hugging Face Spaces), contact me before ordering for a custom quote.
