Akshay Kumar - Applied ML Engineer

About Me

I am a Applied MLE and Computer Vision Engineer with expertise in deploying real-time computer vision models on edge devices, developing ChatBots, and building Retrieval-Augmented Generation (RAG) systems for document-based question answering.

With 1+ years of experience in creating production-ready AI solutions, I specialize in developing high-accuracy computer vision systems (95%+) and working with advanced Large Language Models (LLMs). My primary focus is on enhancing my knowledge and skills in Computer Vision, Natural Language Processing (NLP), and Deep Learning.

Founder of PupsN.com

I am the founder of PupsN.com, a platform dedicated to providing high-quality resources and tools for pet lovers and professionals. Our mission is to make pet care more accessible and enjoyable through innovative solutions.

Work Experience

Computer Vision Engineer

Loksun.ai, Bengaluru, KA

Jan 2025 – Present

Developed a real-time PPE Detection System with frame dropping and ROI functionality, fine-tuning YOLOv11n on a 50% custom dataset to achieve mAP50: 0.8718 and mAP50-95: 0.5999, improving detection accuracy from 70% to ~95%.
Developed a multi-source facial recognition system with over 95% detection accuracy. Optimized model usage for security monitoring and access control by enabling cross-profile face matching, storing detection results in an SQL database, and tracking individual sightings effectively.
Enhanced voice-to-person recognition through optimized signal processing and developed an AI agent using Bedrock Claude for automated blog creation with SEO and image generation.
Created a Retrieval-Augmented Generation (RAG) system using the Gemini model and developed AI agents for email generation and document processing.

AI/ML Intern

Techdome Solutions, Indore, MP

Oct 2023 – Feb 2024

Developed and deployed Catalyst, an AzureEdge application for video feed and sensor data processing, using YOLO for detection (mAP: 0.72) and SORT for tracking, integrated with Kafka and HiveMQ (MQTT).
Built ChatWithData, a chatbot for natural language queries providing data insights as graphs, text summaries, and visualizations.
Developed a Resume Processer using RAG to extract and store relevant data from resumes.
Worked on multiple projects with Generative AI and Transformer NLP models.

Assoc. Solution Developer

Quadwave Consulting, Pune, MH

Aug 2022 – Jan 2023

Worked on Econsys V4, a business management system for specialist contracting firms.
Developed backend services using Java (Liferay, JSP, Servlets, and JBPM) and assisted in unit testing & module implementation.

Education

Bachelor in Computer Application and Cyber Security (BCACS)

Jharkhand Raksha Shakti University, Ranchi, JH

2020 - 2023

CGPA: 8.41

Technical Skills

Machine Learning & AI

Computer Vision NLP Deep Learning Machine Learning LLMs RAG

Frameworks & Libraries

TensorFlow OpenCV YOLO scikit-learn Pandas NumPy Hugging Face

Cloud & Infrastructure

AWS (Bedrock, Lambda) Azure (AzureEdge) Docker MLOps

Data Engineering & Development

SQL NoSQL Kafka MQTT Git FastAPI Flask CI/CD Linux

Projects

Traffic Monitoring

A system to monitor road traffic using video feeds, leveraging YOLOv8 for object detection and SORT for tracking. The model accurately detects and tracks vehicles in real-time, with EasyOCR for text extraction.

YOLO FaceNet EasyOCR OpenCV SORT

View on GitHub

RAG Model Framework

Open-source Retrieval-Augmented Generation model for document-based question answering. Supports multiple file formats and leverages pre-trained models for retrieval, re-ranking and summarization.

Python Transformers Flask RAG

View on GitHub

Test Cheating Detection

Real-time cheating detection system using computer vision. Tracks head orientation, detects unauthorized devices, identifies individuals, and sends alerts for cheating.

Python YOLO OpenCV MediaPipe

View on GitHub

GitHub Stats

GitHub Activity Overview

Most Used Languages

GitHub Contribution Streak

Contact Me

Have a project in mind? Let's discuss how we can work together.

akshaysatyam2003@gmail.com

linkedin.com/in/akshaysatyam2