Open to Opportunities · Varanasi, India

Rhitwij
Mukharjee

>

Building intelligent systems at Softweb Technology, Kolkata. Specializing in MLOps, Computer Vision, NLP, and deploying real-time ML inference at the edge.

0+

Years at Softweb Technology

~0%

Fatigue Incident Reduction

0%

ETL Failure Rate Cut

0.0/10

B.Tech SGPA, BPUT

// expertise

What I Do

Building production-grade ML systems that solve real industrial problems — from edge deployment to enterprise MLOps.

Machine Learning & MLOps

End-to-end ML pipelines using MLflow, W&B, DVC, and Airflow. Production model deployment with automated monitoring, drift detection, and CI/CD integration.

Computer Vision & Edge AI

Real-time vision systems with YOLO, UNet, and MediaPipe. TensorRT-optimized inference on Raspberry Pi for sub-50ms latency in industrial environments.

NLP & Language Models

Text classification and summarization using BERT/GPT transformers. Safety report analysis, multi-label classification, and domain-specific fine-tuning.

// featured work

Selected Projects

High-impact ML systems deployed in production environments across fleet management, mining, and data engineering.

Computer Vision · Edge AI🚛

Fleet Driver Fatigue Detection

Real-time driver drowsiness detection system using MediaPipe face mesh + LSTM temporal analysis. Deployed across a fleet with live MQTT alerts, reducing fatigue incidents by ~30%.

~30% reduction in fatigue incidents
PythonMediaPipeLSTMMQTTRTSPRaspberry Pi
// private repo
MLOps · Edge Deployment⛏️

Mining Anomaly Detection

On-edge TensorRT inference engine for real-time anomaly detection in mining operations. Deployed on Raspberry Pi, processing RTSP camera feeds with <50ms latency.

Sub-50ms edge inference latency
TensorRTYOLOUNetRaspberry PiRTSPPython
// private repo
Data Engineering · MLOps🏗️

ETL Reliability Platform

Robust Airflow-based ETL orchestration with DVC data versioning, Feast feature store, and automated alerting. Cut pipeline failure rate by 40% across multi-source ingestion.

40% fewer pipeline failures
AirflowDVCFeastKafkaDockerAWS S3
// private repo

"The best ML model is the one that reliably runs in production, not the one that wins the benchmark."

— Rhitwij Mukharjee, Data Scientist

// LET'S BUILD SOMETHING

Have a challenging ML problem?
Let's solve it together.

Available for full-time roles, consulting projects, and open-source collaborations in Data Science, MLOps, and Edge AI.