ML Engineer · Chennai, India

Venkatesh
P

RAG Systems & Vector Search

I build production-grade ML systems — end-to-end, from data pipelines to deployed products. Currently building PaperLens, a semantic search engine over arXiv papers.

RAG Pipelines FAISS Vector Search Groq LLaMA-3 FastAPI Docker GCP

View Projects → ↓ Download Resume LinkedIn

Selected Projects

★ Flagship 001

PaperLens

Semantic search engine over arXiv research papers — search by meaning, not keywords. Built end-to-end: arXiv ingestion pipeline → sentence-transformers embeddings (384-dim) → FAISS IVFFlat vector index → FastAPI backend with LRU caching → Groq LLaMA-3.3-70b RAG pipeline for structured AI summaries → Streamlit dashboard.

0.02msp50 latency

76,540queries/sec

1,738×cache speedup

~1.2sAI summary

sentence-transformers FAISS IVFFlat FastAPI Groq LLaMA-3 ChromaDB Streamlit Docker SQLite

↗ GitHub Live Demo — Soon

RAG System 002

Enterprise Ticket Analysis Bot

Production RAG system over 70MB+ enterprise ticket logs. Full pipeline: PDF parsing → chunking → SentenceTransformer embeddings → ChromaDB vector store → Groq Llama-4 Scout for contextual QA. Exposed via FastAPI with Streamlit dashboard.

ChromaDBGroq Llama-4 SentenceTransformersFastAPIStreamlit

↗ GitHub

LLM + RAG 003

Well.AI — Medical LLM Assistant

Medical RAG pipeline with structured retrieval, chunking, and dense embeddings for domain-specific health recommendations. FastAPI microservices deployed on GCP with Supabase session logging for full traceability.

LangChainHuggingFace FastAPIGCPSupabaseDocker

↗ GitHub

Computer Vision 004

Brain Tumor Classifier

CNN achieving 95% accuracy on MRI brain tumor classification across 4 classes. Preprocessing pipeline with augmentation and normalization, containerised clinical inference interface with Docker.

TensorFlowCNN Transfer LearningDocker

↗ GitHub

Tech Stack

Core Competencies

LLM / GenAI

RAG PipelinesFAISS ChromaDBLangChain sentence-transformersGroq API LoRA / QLoRAHuggingFace Prompt Engineering

ML / Deep Learning

PyTorchTensorFlow scikit-learnXGBoost CNNsTransformers Fine-tuningMLflow

Backend & Deployment

FastAPIDocker Docker ComposeGitHub Actions GCPAWS EC2/S3 StreamlitHuggingFace Spaces

Languages & Data

PythonSQL JavaScriptBash PandasNumPy PySparkPostgreSQL

Credentials

Education

Master of Computer Applications — University of Madras, 2023–2025. GPA: 7.3/10

Certifications

AWS Data Engineering Bootcamp Udemy · 2024

Machine Learning A–Z Udemy · 2024

Deep Learning Specialization Coursera · In Progress

Work Experience

Jan 2024 — Apr 2024

INTERNSHIP

AI/ML Engineer Intern

NASO Technologies

Built production RAG pipeline (Well.AI) using LangChain + HuggingFace + ChromaDB; deployed via FastAPI on GCP with Supabase session logging.
Containerised all services with Docker; standardised multi-service setup reducing deployment time significantly.
Designed prompt-engineering workflows for domain-specific health reasoning, improving LLM response quality.

Apr 2024 — Jul 2024

RESEARCH

Machine Learning Researcher

RUSA Project · University of Madras

Engineered SNP-derived genomic features from GWAS data; trained ensemble ML models achieving 87.6% accuracy on Multiple Sclerosis risk prediction.
Built Streamlit dashboard for polygenic risk scoring, SNP visualization, and model interpretability used by the research team.
Deployed reproducible containerised ML workflows on GCP with DVC for data versioning.