Harsh · Python / ML Engineer · 3+ yrs
JuniorBuilt for
About Harsh
Harsh is a Python Developer with 3y. Portfolio: Proof of Concept for Image Similarity (jewelry — Python + TensorFlow + Keras + VGG16), iEra (AI/LLM-driven vector store + content-based chat assistant on Python + React + AI/ML), Customer Purchase Propensity (end-to-end e-commerce model on Python + Pandas + NumPy + Scikit-learn), and…
3+ years of commercial experience in
Skills(32)
Why hire Harsh?
Predictive ML models for agricultural optimization
Functional eCommerce purchase propensity model
Core API functionalities for AI vector store
Project highlights(4)
Proof of Concept for Image Similarity Search for Jewelry – Developer
- Developed a proof of concept for an image similarity search application, specifically for jewelry ring images.
- The application allows users to input any ring photo and receive similar ring images as output.
- Developed the core logic for the image similarity search application.
- Utilized TensorFlow and Keras applications (VGG16, ResNet50) for effective feature extraction from images.
- Implemented cosine similarity using Scikit-learn to compare and measure similarity between extracted image features.
- Developed a user interface using Gradio to enable easy input of ring images and viewing of results.
Key outcomes:
Successfully developed a functional proof of concept for image similarity search.
Enabled user input of jewelry images to retrieve similar results, demonstrating practical application of ML.
iEra – Developer
- Contributed to the development of an AI/LLM-driven vector store and a content-based chat assistant.
- Focused on enhancing similarity searches and real-time user interactions.
- Modified and developed APIs for the iEra Vector Store, enabling JSON input and document management.
- Integrated Scikit-learn and pgvector options into assistant settings for efficient vector store operations.
- Created and updated vector store configuration APIs.
- Developed a content-based chat assistant utilizing Amazon Bedrock for enhanced similarity search capabilities.
- Implemented a Chat Widget with Socket Integration to facilitate real-time interactions.
- Created a Dockerfile for the iEra Vector Store, supporting containerization and deployment.
Key outcomes:
Developed core API functionalities for an AI vector store, handling JSON input and document management.
Successfully integrated Amazon Bedrock to power a content-based chat assistant with advanced similarity searches.
Implemented a real-time chat widget using Socket Integration, improving user experience.
Customer Purchase Propensity Model for eCommerce Using Machine Learning – Data Engineer, Data Scientist, Machine Learning Engineer, Data Analyst
- Developed an end-to-end customer purchase propensity model specifically for eCommerce applications.
- Leveraged machine learning to predict customer buying behavior.
- As Data Engineer, collected, cleaned, and preprocessed raw data for model input.
- As Data Scientist, engineered features, split data into training/testing sets, and standardized features.
- As Machine Learning Engineer, trained and evaluated the model, and tuned hyperparameters for optimal performance.
- As Data Analyst, predicted purchase propensity, generated results tables, and visualized outcomes for stakeholders.
Key outcomes:
Successfully developed and delivered a functional eCommerce purchase propensity model.
Managed the full lifecycle of an ML project, from data preparation to model deployment and result visualization.
Provided actionable insights through predictions and visualizations.
ISRO, Ahmedabad (NAHEP- AAU) – Researcher
- Developed advanced predictive machine learning models for agricultural crop price, acreage, and production optimization.
- Utilized geospatial data and tools like Google Earth Engine to improve yield forecasts and resource allocation.
- Developed predictive ML models using R, Python, and Google Earth Engine for agri-crop optimization.
- Rigorously trained ML models with comprehensive datasets and advanced algorithms to understand patterns and trends.
- Handled diverse Indian and foreign satellite data (e.g., Sentinel-2, MODIS, RISAT-1A) and climate reanalysis data (e.g., CHIRPS, GLDAS, ERA-5).
- Demonstrated expertise in various remote sensing and GIS software, including ENVI, QGIS, and ArcGIS.
- Employed a meticulous evaluation process with robust performance metrics and validation techniques to assess model effectiveness.
- Made high-accuracy predictions to facilitate data-driven decision-making in agriculture.
- Led model development initiatives for accurate price forecasting using machine learning algorithms.
Key outcomes:
Successfully developed and deployed predictive ML models for agricultural optimization over a 2-year period.
Managed and processed diverse satellite and climate data for model development and evaluation.
Led initiatives for accurate price forecasting, demonstrating leadership in ML model development.
Industry experience
AgriTech
1 project- •ISRO, Ahmedabad (NAHEP- AAU)— ResearcherR · Python · Google Earth Engine · QGIS +5
E-commerce
2 projects- •iEra— DeveloperPython · React.js · Artificial Intelligence (AI) · Machine Learning (ML) +5
- •Customer Purchase Propensity Model for eCommerce Using Machine Learning— Data Engineer, Data Scientist, Machine Learning Engineer, Data AnalystPython · Pandas · NumPy · Scikit-learn +1
AI / ML Platform
3 projects- •iEra— DeveloperPython · React.js · Artificial Intelligence (AI) · Machine Learning (ML) +5
- •Customer Purchase Propensity Model for eCommerce Using Machine Learning— Data Engineer, Data Scientist, Machine Learning Engineer, Data AnalystPython · Pandas · NumPy · Scikit-learn +1
- •ISRO, Ahmedabad (NAHEP- AAU)— ResearcherR · Python · Google Earth Engine · QGIS +5
At a Glance
Single contract. No agency markup confusion.
Typically responds within 4 business hours.
Vetted by WitaristNot sure if this is the right fit?
Tell us your requirements and we'll match you with the best candidates.
Harsh
Python Developer