Akshay Dwivedi

Engineered a Kubernetes-based ETL pipeline using Argo Workflows, Python, and direct filesystem I/O, reducing processing time from 30 minutes to 90 seconds (20x performance gain) while replacing a costly vendor-managed workflow Developed a production RAG system for investment recommendation reports using Azure OpenAI, Cohere multimodal embeddings, and Elasticsearch hybrid search (BM25 + HNSW) — processing financial PDFs end-to-end with semantic chunking, session-based multi-tenancy, and page-level citation tracking Architected and maintained a production Medallion Architecture data lake (Bronze/Silver/Gold layers) on Azure, delivering clean, validated, and governed datasets to analysts and portfolio managers across 4 investment teams Designed prompt engineering frameworks with team-specific templates, structured variable injection, and guardrail systems — enabling grounded, hallucination-resistant financial Q&A across global equities, US value, US growth, and emerging market debt portfolios Built real-time streaming SSE endpoints and citation-tracking systems returning source-attributed LLM responses with page-level references, improving analyst trust and auditability of AI-generated financial insights

Graduate Teaching Assistant for Course MSIM-6210 Information Visuals and Dashboards for Business • Assist in the review of course materials and content • Assist learners with identifying connections between academic learning and work experiences • Help in live sessions with materials, announcements, exhibits, and demonstrations • Provide support to students, faculty, and staff during live online class sessions • Interact in content-related forums as assigned • Engage learners through TA office hours and periodic engagement activities through the community platform • Provide individualized feedback to 50 learners on data visualization dashboards in PowerBI and Tableau, business analysis, SQL queries, and code reviews

• Ensuring the safety of games by conducting them in an orderly manner and enforcing the laws of the game • Providing training and mentorship to newer officials while presenting them with constructive feedback

Graduate Teaching Assistant for MISM-6205 Data Wrangling in Business. • Led group and individual tutoring sessions to support students in their coursework and exam preparation. • Graded projects for students and shared best practices for the topics along with assisting them to solve complex problems.

• Automated AWS database provisioning and configuration using Terraform modules, cutting deployment time by 30% and ensuring uniform, error-free database setups. • Executed Python-based data analytics projects with Postgres, Looker, Docker, and ECS. Achieved a 20% reduction in response times, enhancing customer satisfaction. • Developed a serverless Slack bot with AWS Lambda functions to pinpoint build stage and codepipeline errors during deployment resulting in accelerated error resolution time by 20% through precise location identification. • Established CDC pipelines with Kafka and Debezium for real-time change tracking across production databases, integrated with Databricks and DBT for multi-layer data modeling • Orchestrated unified batch and streaming data processing on AWS Cloud using technologies such as Amazon EMR, Kinesis, and AWS Glue, ensuring a comprehensive and scalable approach to data analysis and insights.

• Designed and implemented a Medallion Architecture (bronze/silver/gold) for a US-headquartered insurance company, integrating customer databases and reducing downstream data quality issues" • Engineered and crafted 7+ impactful dashboards in Power BI and Tableau,effectively translated complex data into actionable visualizations, enhancing data-driven decision-making processes for multiple stakeholders. • Spearheaded detailed analysis, optimizing data integration workflows, identifying bottlenecks, and driving NoSQL(AWS Dynamo DB) enhancements for a remarkable 25% reduction in data processing time. • Simplified ETL tasks using SSIS, Apache Airflow, Kubernetes, S3 Object Storage, and Python, creating a modern data setup. Cut data processing time from 7 to 3 hours for over 50,000 daily records. • Implemented effective data validation and integrity processes, ensuring accurate data mining outcomes. Achieved a 95% data accuracy rate, improving the reliability of insights and recommendations. • Created a robust monitoring and alerting system using AWS CloudWatch and custom scripts, reducing unplanned downtime by 40% and improving incident response times. • Collaborated with 20+ member of cross-functional teams to create and implement data management policies and best practices, leading to improved data security and compliance.

• Applied comprehensive data analysis to combine multiple data sources and highlight key findings, resulting in a 10% improvement in KPIs. • Executed A/B testing on various visualization formats, resulting in a 15% performance improvement uplift in user engagement and awareness campaigns.