Sandhya Shankar

I am an Aspiring Data Engineer actively seeking better opportunities.

Data engineering isn't just a job for me — it's a space I genuinely love exploring. I work across the modern data stack (Apache Flink, Beam, GCP, Databricks, Kafka) and have a habit of investing time in research to bring fresh thinking to real problems at work. From building scalable ETL frameworks to evaluating LLMs for production use, I'm energized by work that sits at the edge of what's possible with data.

Data Engineer - Associate at FedEx (2026-02 – Present)

Building Python-based frameworks for ETL automation and streamlining scalable GCP data processing workflows.

• Built a Python-based framework to convert SQL and config files into Apache Beam YAML pipelines, enabling automated Google Cloud Dataflow ETL jobs on BigQuery.
• Implemented SQL parsing, YAML pipeline generation, validation, and Cloud Storage staging to streamline scalable GCP data processing workflows.

Junior Associate Engineer at FedEx (2025-08 – 2026-02)

Designed and developed real-time data products using Apache Flink SQL and Databricks pipelines for enterprise logistics systems.

• Designed and developed real-time data products using Apache Flink SQL to pipeline package tracking and shipping events across enterprise logistics systems.
• Implemented JSON-based schema mappings and transformations in Flink SQL to standardize critical fields.
• Built Databricks ingestion pipelines to process raw package events from Confluent Kafka topics, transforming them into structured DataFrames for Testing, downstream analytics and Power BI reporting.

Data Engineer Intern at FedEx (2025-01 – 2025-08)

Optimized OCR extraction accuracy and developed rule-based evaluations for the iClear Project.

• Optimized OCR Extraction Accuracy of LLM (>90%) through Prompt Engineering, multilingual field descriptions, and comparative analysis, improving extraction across 48+ documents for the iClear Project.
• Developed and tested rule-based evaluations, including Rule Engine Testing, Pipeline Data Flow Validation, and Undetermined Field Checks for structured data compliance.

Machine Learning Intern at Adani AI Labs (2024-06 – 2024-07)

Developed time-series forecasting models for power demand and price prediction and created PowerBI dashboards.

• Developed time-series forecasting models (LSTM, XGBoost, ARIMA, SARIMAX) for power demand and price prediction for Power Optimization Project by AEML.
• Engineered lag and Seasonality features, improving baseline forecasting accuracy by ~10%.
• Developed PowerBI Dashboard for Forecasted Results for multiple use-cases as visual representation to vendors and business-side.
• Dealt with web-crawling, data preprocessing, validation, feature addition etc.

GenAI Research Intern at Petavue (2024-01 – 2024-04)

Evaluated LLMs for natural-language-to-SQL generation and built evaluation frameworks for performance analysis.

• Evaluated LLMs (LLaMA, Mixtral, CodeLlama, SQLCoder, GPT-4, Claude) for natural-language-to-SQL generation using the Spider benchmark dataset.
• Designed and benchmarked zero-shot and few-shot prompting strategies, analyzing performance across SQL query complexity levels.
• Built evaluation frameworks measuring accuracy, latency, throughput, and inference cost across Bedrock, Anyscale, and self-hosted GPU environments.

B. Tech in Information Technology – College Of Engineering, Guindy (Anna University) (2021-11 – 2025-05)