Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI/ML Engineer with 5 years of experience building production machine learning and NLP systems in Python and SQL. I specialize in LLM applications such as Retrieval-Augmented Generation (RAG), semantic search, embeddings, and LLM evaluation, deploying solutions as containerized microservices on Google Cloud. I have a proven track record of shipping predictive models, Transformer-based NLP, and scalable data pipelines that improve operational efficiency and decision-making. I enjoy building end-to-end ML systems—from data validation and feature engineering to monitoring and iteration based on measurable evaluation results.…AI/ML Engineer with 5 years of experience building production machine learning and NLP systems in Python and SQL. I specialize in LLM applications such as Retrieval-Augmented Generation (RAG), semantic search, embeddings, and LLM evaluation, deploying solutions as containerized microservices on Google Cloud. I have a proven track record of shipping predictive models, Transformer-based NLP, and scalable data pipelines that improve operational efficiency and decision-making. I enjoy building end-to-end ML systems—from data validation and feature engineering to monitoring and iteration based on measurable evaluation results.

Aryan Sharma

AI Engineer, AI Developer, Data Scientist, +3





AI/ML Engineer with 5 years of experience building production machine learning and NLP systems in Python and SQL. I specialize in LLM applications such as Retrieval-Augmented Generation (RAG), semantic search, embeddings, and LLM evaluation, deploying solutions as containerized microservices on Google Cloud. I have a proven track record of shipping predictive models, Transformer-based NLP, and scalable data pipelines that improve operational efficiency and decision-making. I enjoy building end-to-end ML systems—from data validation and feature engineering to monitoring and iteration based on measurable evaluation results.…AI/ML Engineer with 5 years of experience building production machine learning and NLP systems in Python and SQL. I specialize in LLM applications such as Retrieval-Augmented Generation (RAG), semantic search, embeddings, and LLM evaluation, deploying solutions as containerized microservices on Google Cloud. I have a proven track record of shipping predictive models, Transformer-based NLP, and scalable data pipelines that improve operational efficiency and decision-making. I enjoy building end-to-end ML systems—from data validation and feature engineering to monitoring and iteration based on measurable evaluation results.

Available to hire

AI/ML Engineer with 5 years of experience building production machine learning and NLP systems in Python and SQL. I specialize in LLM applications such as Retrieval-Augmented Generation (RAG), semantic search, embeddings, and LLM evaluation, deploying solutions as containerized microservices on Google Cloud.

I have a proven track record of shipping predictive models, Transformer-based NLP, and scalable data pipelines that improve operational efficiency and decision-making. I enjoy building end-to-end ML systems—from data validation and feature engineering to monitoring and iteration based on measurable evaluation results.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Advanced

Hindi

Advanced

Work Experience

AI Engineer at California State University, Library & Research Systems

August 1, 2024 - December 31, 2025

Built an LLM-powered research assistant using Retrieval-Augmented Generation (RAG) over 10K+ academic resources with semantic search and embeddings, reducing average resource-discovery time by ~30%. Designed document ingestion and chunking pipelines in Python for large structured/unstructured research data (PDFs, metadata, usage logs) to improve retrieval relevance via iterative prompt engineering and evaluation. Established an LLM evaluation framework tracking response latency, answer accuracy, task completion rate, and failure rate; tuned retrieval parameters to reduce hallucination-related failures by ~25%. Deployed containerized inference services with Docker and FastAPI on Google Cloud Platform (Cloud Run, Cloud Storage) exposing REST APIs for a library search interface serving 5K+ monthly users. Implemented data validation and preprocessing pipelines using Pandas and SQL/PostgreSQL to improve data quality and reliability for downstream analytics and ML workloads.

Data Scientist at Sabudh Foundation

July 1, 2021 - June 30, 2023

Delivered end-to-end ML solutions in Python and SQL—from problem framing and feature engineering through training, evaluation, and deployment—to support business and operational decision-making. Built and productionized predictive models (gradient boosting, regression, time-series forecasting) with Scikit-learn for demand forecasting and program planning, improving forecast accuracy by ~18% versus baseline heuristics. Designed and optimized batch data pipelines for large datasets (millions of records), reducing reporting and batch processing time by 40% through query optimization and workflow automation. Applied Transformer-based NLP (BERT via Hugging Face) for text classification and information extraction on unstructured program data, automating manual review. Implemented automated data preprocessing, validation, and drift checks to improve data reliability across retraining cycles. Mentored junior trainees and collaborated with product, engineering, and business stakeholders to

Data Scientist at Sabudh Foundation, Mohali, India

July 1, 2021 - June 30, 2023

Delivered end-to-end machine learning solutions in Python and SQL from problem framing and feature engineering through training, evaluation, and deployment to support business and operational decision-making. Built and productionized predictive models (gradient boosting, regression, time-series forecasting) with Scikit-learn for demand forecasting and program planning, improving forecast accuracy by ~18% over baseline heuristics. Designed and optimized batch data pipelines over millions of records, reducing reporting and batch processing time by 40% via query optimization and workflow automation. Applied Transformer-based NLP models (BERT via Hugging Face) for text classification and information extraction on unstructured program data, automating previously manual review. Implemented automated data preprocessing/validation and drift checks to improve data reliability and downstream model performance across retraining cycles.

Data Scientist at Zoho

January 1, 2020 - June 30, 2021

Developed NLP models for customer-support text analytics; fine-tuned LSTM and BERT architectures with TensorFlow, PyTorch, and Hugging Face Transformers on Databricks, improving contextual sentiment classification accuracy by 20%. Engineered an intelligent support assistant integrated with Salesforce and Zendesk via REST APIs for real-time CRM data retrieval and updates, reducing average support resolution time by ~20%. Shipped multi-channel conversational interfaces (web, mobile, WhatsApp) using Flask/FastAPI and Dialogflow orchestrated with scheduled Databricks Jobs, contributing to a 35% lift in customer engagement. Owned the full model lifecycle including data preparation, training, evaluation, versioned deployment, and monitoring—establishing early MLOps practices (Git workflows, reproducible pipelines, containerized services) adopted by the team.

Data Scientist at Zoho, India

January 1, 2020 - June 30, 2021

Developed NLP models for customer-support text analytics, fine-tuning LSTM and BERT architectures using TensorFlow/PyTorch and Hugging Face Transformers on Databricks to improve contextual sentiment classification accuracy by 20%. Engineered an intelligent support assistant integrated with Salesforce and Zendesk via REST APIs for real-time CRM data retrieval and updates, reducing average support resolution time by ~20%. Shipped multi-channel conversational interfaces (web, mobile, WhatsApp) using Flask/FastAPI and Dialogflow orchestrated through scheduled Databricks Jobs, contributing to a 35% lift in customer engagement. Owned the model lifecycle (data prep, training, evaluation, versioned deployment, monitoring) and established early MLOps practices including Git-based workflows, reproducible pipelines, and containerized services.