Home›LinkedIn
Building in Public

LinkedIn Posts

I share what I learn as I build: RAG systems, LLM orchestration, ML experiments, career reflections, and the occasional late-night AI rabbit hole.

Follow on LinkedIn
Activity
Posts Shared16+
Topics Covered7
Profile @bijaya-kumar-pariyar
Video Post

My Most-Viewed Post

Tap to watch the full video directly on LinkedIn

Modular RAG API with FastAPI & Groq

My most-viewed post: a walkthrough of a configurable RAG system with switchable vector backends, multiple chunking strategies, and LLM-powered, context-grounded generation.

Watch on LinkedIn
Post Feed

Recent Posts

Click any card to read the full post directly on LinkedIn

Improving RAG Precision with Reranking

Why retrieval alone isn't enough: two-stage retrieval + cross-encoder reranking, stress-tested with hard negatives. Better Hit Rate@k and MRR.

#RAG#Reranking
Read on LinkedIn
Text Chunking Strategies for RAG

Comparing fixed-size, sentence, recursive, and semantic chunking, and how each one affects RAG retrieval quality.

#RAG#NLP
Read on LinkedIn
SHAP-Based Churn Prediction

My research preprint: SHAP for transparent feature importance plus iterative hyperparameter tuning on telecom churn data.

#SHAP#Research
Read on LinkedIn
Building a RAG Pipeline with LLMs

A hands-on RAG system grounding LLM responses in retrieved context using sentence-transformers, Qdrant, and Groq.

#RAG#LLMs
Read on LinkedIn
Semantic Search with FAISS & FastAPI

A lightweight semantic search workflow with FAISS and sentence-transformers, served via FastAPI with live index rebuilding.

#SemanticSearch#FAISS
Read on LinkedIn
Text Embeddings Explorer CLI

An open-source CLI to explore embeddings across MiniLM, MPNet, BGE, and E5, with similarity comparisons and PCA visualizations.

#Embeddings#NLP
Read on LinkedIn
PM2.5 Air-Quality Anomaly Detection

Comparing Z-Score, IQR, and rolling-statistics methods to detect outliers in Beijing air-quality time-series data.

#AnomalyDetection#TimeSeries
Read on LinkedIn
ML-Driven Retail Price Optimization

An ML pricing workflow with Optuna tuning, SHAP explainability, and profit-optimization logic for retail decisions.

#MachineLearning#Optuna
Read on LinkedIn
Price Elasticity & Competitive Pricing

Random Forest modeling with price-elasticity analysis and competitive pricing simulation to optimize retail margins.

#PriceOptimization#DataScience
Read on LinkedIn
Clustering-Based Anomaly Detection

KMeans and DBSCAN anomaly detection with clear visualizations on real-world data, part of my anomaly-detection series.

#AnomalyDetection#Clustering
Read on LinkedIn
Customer Segmentation with KMeans

KMeans clustering and RFM analysis to segment retail customers from two years of transactions for targeted marketing.

#Clustering#RetailAnalytics
Read on LinkedIn
See all activity on LinkedIn
More Posts

Also on LinkedIn

Click any card to read the full post directly on LinkedIn

Anomaly Detection & Fraud

Detecting anomalies in fraud datasets: methods, metrics, and what actually works.

#MachineLearning #FraudDetection
Read on LinkedIn
Data Science & Anomaly Detection

How I approach anomaly detection in large-scale datasets with multiple algorithm comparison.

#DataScience #Anomaly
Read on LinkedIn
IEEE Fraud Detection & EDA

Classification on 1M+ IEEE-CIS transactions: EDA, feature engineering, and XGBoost results.

#Classification #IEEE
Read on LinkedIn
Time Series Forecasting

Forecasting supermarket prices across 1.7M rows: LightGBM, Prophet, and the features that matter.

#TimeSeries #LightGBM
Read on LinkedIn

Let's Connect & Grow Together

I post regularly about AI engineering, RAG systems, ML insights, and lessons from building production systems. Join the conversation!

Follow on LinkedIn See My Projects