About Me

AWS Machine Learning Associate Certified AI/ML Engineer specializing in deploying production LLM systems and scalable ML infrastructure using Amazon Bedrock and SageMaker. Proven expertise in prompt engineering, RAG architectures, and multi-model LLM orchestration with hands-on experience achieving 60% efficiency gains and 92% accuracy in production AI systems. Strong background in ML governance, model auditability, and end-to-end deployment of compliant, enterprise-grade AI solutions. Published researcher ready to drive ML innovation at technology companies.

Harsh Shroff

Publications

FilletCam AI: Precision fillet color profiling using deep learning. Journal of Agriculture and Food Research (2024).
What’s unique: First real-time edge deployment for aquaculture color grading (92% accuracy).
DOI: 10.1016/j.jafr.2024.101461

Mosquito identification using ML on embedded systems. IEEE ITU Kaleidoscope Conference (2021).
What’s unique: Embedded ML on ARM Cortex-M; low-latency ( <100 ms) inference with high accuracy.
DOI: 10.23919/ITUK53220.2021.9662116

For Recruiters

  • Open to: Machine Learning Engineer, Production ML Systems, LLM Applications
  • Relocation: Open to relocation
  • Start date: Immediately available

Technical Skills

Machine Learning & AI

Generative AI, LLMs, RAG, LangChain, PyTorch, TensorFlow, Computer Vision, YOLOv8, OpenCV, NLP, Scikit-learn, Keras

Deep Learning Specializations

Neural Networks, CNNs, RNNs, Transformers, Fine-tuning, RLHF, Anomaly Detection, Reinforcement Learning, Time Series

Software Engineering

Python, C++, Java, SQL, FastAPI, Flask, Git, Docker, Kubernetes, CI/CD, API Development, Microservices, Test-Driven Development

Cloud & MLOps

AWS (Bedrock, SageMaker, Lambda, Textract, EC2, S3), MLflow, Weights & Biases, Model Monitoring, Model Governance, A/B Testing, Auditability

Data Engineering

Pandas, NumPy, Spark, Hadoop, Feature Engineering, Data Pipelines, ETL, PostgreSQL, MongoDB, Redis

Deployment & Tools

Streamlit, Gradio, Jupyter, CUDA, Edge Computing, NVIDIA Jetson, Model Serving, Performance Optimization

Core Skills Proficiency

Based on production experience and project implementations

Machine Learning & AI

AWS Bedrock & SageMaker Expert
LLMs & RAG Architectures Expert
PyTorch & TensorFlow Advanced
Computer Vision (YOLOv8, OpenCV) Advanced

Cloud & MLOps

AWS Services (Lambda, S3, EC2) Expert
MLflow & Model Monitoring Advanced
Docker & Kubernetes Advanced

Programming & Engineering

Python Expert
FastAPI & Flask Advanced
SQL & NoSQL Databases Advanced
C++ & Java Intermediate

Data Engineering

Pandas & NumPy Expert
ETL Pipelines & Data Processing Advanced
Spark & Hadoop Intermediate

Professional Experience

Mar 2023 - Present

AI/ML Researcher

UMBC Center for Real-time Distributed Sensing and Autonomy

PyTorch Computer Vision Edge Computing Autonomous Systems Python
  • Led 5-engineer team developing distributed ML systems for multi-modal sensor data fusion under US Army Research Lab funding, achieving 20% performance improvement in real-time optimization systems.
  • Designed production monitoring dashboards and real-time inference pipelines enabling autonomous systems deployment with computer vision and sensor fusion capabilities. Demo Video
  • Engineered production-ready multi-modal ML systems integrating advanced perception algorithms for next-generation AI applications using deep learning frameworks and edge computing platforms.
  • Built CHARLIE Voice Assistant, a voice-enabled AI agent integrating speech recognition, LLM inference, and real-time dialogue management for autonomous system interfaces.
Jun 2024 - May 2025

ML Engineer - Production AI Systems (Contract)

VITG Corp., Halethorpe, MD

AWS Bedrock SageMaker RAG Textract Python PostgreSQL
  • Architected and deployed production LLM automation system using AWS Bedrock and SageMaker, leveraging Claude 3.5 and Llama models to reduce candidate screening time by 60% for 200+ employee organization, processing 2,500+ regulatory documents with 40% efficiency improvement.
  • Implemented RAG-based conversational AI chatbot with retrieval-augmented generation architecture using AWS Textract for document ingestion, vector embeddings for semantic search, and prompt-engineered LLM responses, achieving enterprise-grade compliance and auditability for regulatory workflows.
  • Established ML governance framework and model monitoring pipelines ensuring compliance, auditability, and performance tracking across deployed LLM systems with automated logging, versioning, and evaluation metrics.
  • Developed scalable ETL pipelines and RESTful APIs for real-time geospatial data processing and multi-model LLM inference, enabling seamless integration with enterprise systems.
Aug 2023 - Dec 2023

Data Scientist - Computer Vision

The Conservation Fund, Shepherdstown, WV

YOLOv8 OpenCV TensorFlow Edge Deployment Flask
  • Developed production-ready computer vision system achieving 92% accuracy using YOLOv8 and OpenCV, deployed on edge devices for real-time quality assessment with end-to-end ML pipeline from data collection to model serving.
  • Co-authored peer-reviewed research demonstrating commercial impact of deep learning applications, enabling commercial deployment of AI quality control systems with automated reporting capabilities.

Education

Master's in Data Science

University of Maryland Baltimore County (UMBC)

Aug 2022 - May 2024

GPA: 3.8/4.0

Relevant Coursework: Natural Language Processing, Data Analysis & Visualization, Data Management, Big Data Processing, Financial Data Science, Machine Learning, Artificial Intelligence

Bachelor's in Electronics & Communications Engineering

Gujarat Technological University

Jun 2019 - May 2022

GPA: 3.8/4.0

Relevant Coursework: Machine Learning & AI, Python Programming, Internet of Things, Embedded Systems

Key Projects

Enterprise Document Intelligence Platform

AWS, Textract, S3, Lambda, Python, LLM APIs, Streamlit

  • Built scalable document processing system automating extraction from 2,500+ PDFs using LLM-powered OCR pipelines and cloud infrastructure with real-time API endpoints and automated compliance workflows.
  • Implemented interactive dashboards enabling stakeholders to query and visualize insights from unstructured data with automated reporting and decision-support capabilities.

Multi-Modal Recommendation Engine

Python, OpenAI API, Computer Vision, PostgreSQL, A/B Testing

  • Engineered hybrid recommendation system combining structured data analysis with computer vision, reducing user bounce rate by 20% and increasing engagement by 40% through personalized AI-driven recommendations.
  • Implemented scalable data pipelines and A/B testing framework for continuous model improvement and performance monitoring with real-time analytics and user feedback integration.

AI-Powered Medical Diagnosis System

Python, TensorFlow, Computer Vision, Flask, Gradio

  • Designed end-to-end ML web application for medical image classification achieving high diagnostic accuracy using deep learning and computer vision techniques with automated feature extraction.
  • Deployed production-ready inference pipeline with interactive web interface, enabling real-time medical assessment and automated reporting capabilities for healthcare applications.

Real-Time Analytics Dashboard

Python, Dash, Machine Learning, Statistical Analysis, Plotly

  • Built interactive ML-powered analytics platform with real-time data ingestion, model inference, and visualization capabilities for performance optimization and predictive insights.
  • Integrated predictive analytics improving performance outcomes by 20% through data-driven insights, automated reporting, and statistical modeling techniques.

Gesture-Based Control System

Python, OpenCV, Real-time Processing, Edge Computing, NVIDIA Jetson

  • Developed real-time gesture recognition system using computer vision and machine learning for intuitive device control and human-computer interaction with low-latency inference.
  • Implemented efficient edge computing solution enabling responsive control in resource-constrained environments with optimized model deployment on embedded hardware.

Beyond Work

Robotics

Exploring how robots can assist in everyday tasks and specialized environments.

Generative AI

Experimenting with the latest LLMs and generative models to create new possibilities.

Outdoor Activities

Hiking and exploring nature to find balance and inspiration.

Continuous Learning

Always seeking new knowledge and skills in emerging technologies.

Achievements & Certifications

AWS Machine Learning Associate Certification

Valid through Apr 2028

View Credential

AWS Machine Learning Foundations

Core ML concepts & AWS integrations

NVIDIA Deep Learning Institute

Transformer-Based NLP, GPU Computing

2 Published Research Papers

IEEE & Journal of Agriculture

PDF Email LinkedIn