Open to opportunities

Abhishek Puli

@abhishekpuli

ML/GenAI engineer building production-style Hybrid RAG and computer vision systems with FastAPI, Docker, and explainability.

India

Message

What I'm looking for

I’m looking for a role where I can build production-style Hybrid RAG and computer vision systems, deploy APIs with FastAPI and Docker, and add explainability and safeguards—not just prototype models.

I build and ship production-style AI systems—hybrid Retrieval-Augmented Generation and computer vision models—rather than stopping at notebooks.

I delivered deployed applications including a fetal health classifier with ~96% accuracy and ~0.98 ROC-AUC, and a real-time classroom surveillance system using YOLOv11 with 90%+ detection accuracy.

For deployment and transparency, I use FastAPI, Docker, and SHAP, and I design retrieval and prompt safeguards (FAISS, BM25, cross-encoder reranking, and locally hosted Llama 3 via Ollama) to reduce hallucinations and prompt injection.

Experience

Work history, roles, and key accomplishments

Hybrid RAG Academy Assistant

Github

Jun 2026 - Jul 2026 (1 month)

Built a production-style hybrid Retrieval-Augmented Generation (RAG) assistant for domain-specific Q&A using FAISS, BM25, and cross-encoder reranking.

Python fastAPI Streamlit Ollama Faiss BM25 Docker RAG

Fetal Health Classification System Developer

Github

Apr 2026 - Apr 2026 (0 months)

Built an end-to-end fetal health classification system using XGBoost on CTG data, achieving ~96% accuracy and ~0.98 ROC-AUC.

XGBoost SHAP fastAPI Docker Python

Smart Classroom Surveillance Developer

Github

Dec 2024 - Apr 2025 (4 months)

Developed a real-time AI surveillance system using YOLOv11 achieving 90%+ detection accuracy on unauthorized mobile phone usage in classrooms.

YOLOv11 Computer Vision Python

Machine Learning Intern

Indian Statistical Institute (ISI)

May 2024 - Jun 2024 (1 month)

Participated in a hybrid training program focused on applied ML techniques. Collaborated in a team of 3 to analyze real-world restaurant data and built a Linear Regression model.

Python Pandas scikit learn Matplotlib EDA