Prachi · Python / GenAI Engineer · 4.5+ yrs
Mid-LevelAbout Prachi
Data scientist + Generative AI engineer with delivery across LLM-powered document classification, RAG pharma chatbot on GPT-4 + ChromaDB, ML return-volume prediction for logistics, and e-commerce LLM customer support chatbot with intent + NER + sentiment.
Core expertise
Additional skills(23)
FastAPI
Docker
GitHub Actions
scikit-learn
PostgreSQLWhy hire Prachi?
Built LLM-powered document classification + entity extraction pipeline for enterprise documents.
Delivered AI-powered pharma chatbot on GPT-4 + RAG + ChromaDB.
Shipped Return Volume Prediction ML solution for a logistics company.
Built e-commerce LLM chatbot with intent detection, NER, sentiment + feedback loops.
Hands-on GenAI / LLM engineering
RAG + intent + NER + sentiment full-stack ML
Multiple production LLM systems
Project highlights(4)
N23 — Document Classification using LLMs – Generative AI Engineer
- LLM-powered pipeline for document classification and entity extraction in enterprise documents — combining GPT-4 with NER models for intelligent classification and tagging.
- Designed and implemented classification + NER pipeline using GPT-4 and custom-trained models.
- Integrated Label Studio for human-in-the-loop feedback.
- Used AWS Textract for OCR-based content extraction.
- Deployed models via SageMaker, triggered workflows with Lambda, exposed APIs via FastAPI.
- Packaged services via Docker and managed with ECS.
- Maintained CI/CD pipelines via GitHub Actions.
- Tuned prompt templates to improve classification accuracy and reduce LLM token consumption.
Key outcomes:
Reduced LLM token consumption + improved classification accuracy via prompt-template tuning.
Built human-in-the-loop pipeline with Label Studio + GPT-4 + custom NER models.
AI-Powered Pharma Chatbot – Generative AI Engineer
- Generative AI chatbot answering medical and research-based queries using GPT-4 + RAG with ChromaDB. Hosted on AWS ECS + S3.
- Implemented RAG architecture using LangChain for context-rich responses.
- Embedded pharma documents using OpenAI embeddings in ChromaDB.
- Developed FastAPI microservices behind Docker containers.
- Managed vector indexing and retrieval workflows.
- Deployed backend on AWS ECS integrated with S3, RDS, ECR.
- Created prompt templates and adjusted token management to control API costs.
- Collaborated with Svelte JS frontend team for API consumption.
- Implemented logging, monitoring and access control for LLM endpoints.
Key outcomes:
Built RAG-based pharma chatbot with ChromaDB vector store and AWS ECS deployment.
Optimised prompt templates to control API costs.
Return Volume Prediction (RVP) – Data Scientist
- AI solution to forecast return volumes of parcels for a logistics company using ML, integrated with dashboards and APIs.
- Engineered time-series ML models using scikit-learn and XGBoost.
- Built FastAPI APIs to expose predictions to business dashboards.
- Integrated Power BI for visualising return volume trends.
- Processed structured data from PostgreSQL and S3.
- Deployed services using Docker and automated workflows with GitHub Actions.
- Managed data ingestion pipelines using AWS Lambda.
- Performed hyperparameter tuning and model evaluation on validation data.
Key outcomes:
Engineered scikit-learn + XGBoost time-series models for parcel return-volume prediction.
Integrated Power BI for stakeholder-facing return-volume trend visualisation.
E-Commerce LLM Chatbot – AI Engineer
E-commerce LLM chatbot with intent detection, NER, sentiment and feedback loops.
Key outcomes:
Built multi-turn e-commerce LLM chatbot with FAISS semantic search and feedback-loop fine-tuning.
4.5+ years of industry experience
- •N23 — Document Classification using LLMs— Generative AI EngineerGPT-4 · FastAPI · AWS Textract · SageMaker +5
- •AI-Powered Pharma Chatbot— Generative AI EngineerGPT-4 · LangChain · ChromaDB · OpenAI +6
- •Return Volume Prediction (RVP)— Data Scientistscikit-learn · XGBoost · FastAPI · Power BI +5
- •E-Commerce LLM Chatbot— AI EngineerLangChain · GPT-4 · FAISS · FastAPI +4
- •Return Volume Prediction (RVP)— Data Scientistscikit-learn · XGBoost · FastAPI · Power BI +5
At a Glance
Single contract. Billed in USD.
Typically responds within 4 business hours.
Top Skills
Vetted by WitaristNot sure if this is the right fit?
Tell us your requirements and we'll match you with the best candidates.
Prachi
Data Scientist