I am Anshuman Sharma

I am a Data Professional with an MS in Data Science with around 2 years of industry experience specializing in Data Engineering, Data Science and Data Analysis. I am Proficient in ML, Statistics, and adept at building and integrating data solutions, including real-time analytics and reporting. Lean Six Sigma Green Belt certified, with a track record of driving efficiency and quality improvement initiatives.

Professional Experience

Visceral.ai : Machine Learning Engineer

Feb, 2024 - Present

• Implemented an LLM auto-parser using Hermes Theta2 (LLaMA 3) on Modal, reducing survey building time by 95%.
• Developed an application for report generation across the platform, utilizing FastAPI to interface seamlessly with Hermes.
• Engineered automated plot generation to visualize underlying data for analysis and presentation.
• Designed and deployed a Retrieval-Augmented Generation (RAG) system using LLaMA 3.1, containerizing it with Docker for efficient deployment on client systems.
• Created a Chroma DB database to support the RAG system, incorporating annotations to improve response accuracy and context.
• Designed and developed consumer data access, deletion, and opt-out mechanisms, ensuring adherence to CCPA and GDPR requirements.
• Established a PostgreSQL database to initiate data collection for the Reinforcement Learning with Human Feedback (RLHF) pipeline.
• Set up evaluation metrics for Hermes Theta2, monitoring improvements in our RLHF-enhanced version of the model.
• Implemented and deployed an auto-classifier API on AWS Lambda, enabling scalable and efficient classification.

Research Assistant (UCI)

Feb, 2024 - Present

• Conducting research in data-intensive computing for collaborative data analytics at scale.
• Working closely with the research team at Donald Bren Hall to develop and implement web-based interactive workflows
• Collaborating with team members to analyze data and draw meaningful insights.

City of Irvine - Machine Learning Engineer for Capstone Project

Sep, 2023 - Dec,2023

• Developed an LLM-powered chatbot using Flask, React, and Llama2, achieving a 97% reduction in time to find contract details.
• Optimized a 1M+ document dataset with the Llama Index and fine-tuned the Llama2 model via prompt engineering.
• Dockerized the application for consistency and portability, and orchestrated its deployment using Kubernetes.
• Architected a CI/CD pipeline with monitoring and auto-retraining for long-term Chatbot efficacy.

Dell Technologies - Data Engineer Intern

Jun, 2023 - Aug,2023

• Uncovered manual reporting inefficiencies via 50+ stakeholder interviews. Derived actionable KPIs for performance improvement.
• Streamlined reporting for Marketing team (FMMs), saving over 5,000 hours annually, and an additional 10,000 hours in the PAN ISG initiative.
• Developed and deployed a data integration framework that consolidated data from 6+ sources into a warehouse (Azure Synapse Analytics).
• Spearheaded the integration of Apache Airflow, enabling automated and scalable data orchestration for the data platform.

Accenture - Data Engineer

Oct, 2020 - Jul, 2022

Electrolux E-commerce Search Optimization

• Developed a hybrid recommendation system using collaborative filtering and content-based filtering, leading to a 12% conversion lift.
• Built real-time product search pipelines in Python with Apache Spark for efficient processing, enabling up-to-date search results and improved user experience.
• Built data pipelines on AWS Glue to ingest & transform millions of daily customer events (purchases, demographics, browsing), enabling real-time recommendations & search filtering.
• Implemented Airflow logging, achieving 20% faster troubleshooting and error identification for data pipelines.

Data Modeling and Optimization

• Designed dimensional data models in SAP BW on HANA, integrating customer & sales data for efficient analysis.
• Optimized data pipelines, achieving a 20% reduction in North American data transfer delays.
• Led a team of 5 to achieve a 98% on-time delivery rate for critical BI reports, exceeding SLA benchmarks.
• Owned the data pipeline lifecycle, ensuring rigorous testing, zero-downtime deployments, and timely delivery of critical BI reports.

Samsung R&D - Data Engineer Intern

Feb, 2020 - April, 2020

• Automated 90% of manual TV testing workload with Robot Framework, saving an estimated 200 man-hours per week.
• Built a real-time data ingestion system using Apache Kafka to capture and process test logs, enabling the generation of reports with Grafana.