Pranav Shridhar Kowadkar

Leading research on fine-tuning language models for neuroimaging classification tasks, with a focus on improving diagnostic accuracy for neurological disorders.

• Developed a novel approach to feature extraction from MRI scans using transformer architectures
• Implemented distributed training pipelines that reduced model training time by 60%
• Collaborated with neuroscience researchers to validate model outputs against clinical diagnoses
• Published findings in peer-reviewed journals and presented at international conferences

Enhanced financial data processing and system performance through advanced scripting and database monitoring.

• Strengthened Oracle-driven data pipelines using PL/SQL and Unix scripting, boosting query speeds by 20%
• Implemented Prometheus metrics for database monitoring and built custom dashboards in Grafana
• Enhanced Python code quality with targeted refactoring for improved deployment stability
• Developed an observability framework for AWS Lambda functions, enabling proactive system troubleshooting

Improved data processing efficiency and analytics through advanced SQL, Python, and pipeline automation.

• Implemented advanced SQL queries and optimized Python data manipulation for processing over 1M data points
• Designed and deployed Airflow-based data pipelines on AWS EC2, reducing manual efforts by 40%
• Championed the adoption of Airflow by developing documentation and training sessions for team members
• Created interactive Looker Studio dashboards to drive data-informed strategic decisions

Developed robust data pipelines and dashboards to support targeted marketing efforts and real-time analytics.

• Architected Databricks pipelines using Spark and Python to process Snowflake healthcare data
• Developed PowerBI dashboards to visualize consumer behavior metrics, enhancing campaign efficiency
• Optimized ETL processes for real-time data ingestion using Azure Data Factory and Python
• Created automated data quality monitoring systems to ensure compliance with privacy regulations

Spearheaded the modernization of CAD/CAM applications and automated testing frameworks to enhance system performance.

• Fixed critical memory issues in CATIA applications using C/C++ and memory profiling tools
• Led the migration of CATIA applications to a Linux environment, reducing downtime by 40%
• Automated CADAM testing with Python, reducing execution time by over 87%
• Developed interactive dashboards for self-service reporting and team performance analytics

Enhanced aerospace design and simulation processes through analytical modeling and automation.

• Boosted VTOL UAV design efficiency by 17% using CFD simulations and Python statistical analysis
• Developed interactive Tableau dashboards for real-time propulsion modeling
• Optimized propeller performance with CATIA V5 and 3D scan correlations
• Automated aerospace calculations with Python, reducing report generation time dramatically

Modernized legacy banking systems and optimized application performance for robust financial operations.

• Developed a C# transaction processing engine integrated with SQL Server, reducing manual reconciliation by 75%
• Optimized financial application performance using OOP principles and multithreading
• Redesigned SQL stored procedures and indexing strategies to cut query latency by 35%
• Implemented automated reporting solutions to transform raw data into actionable business intelligence

Completed a comprehensive program focused on advanced data science techniques, machine learning algorithms, and big data technologies.

Developed a strong foundation in mechanical engineering principles while exploring interdisciplinary applications in aerodynamics and computational design.

• Implemented collaborative filtering and content-based recommendation algorithms
• Designed a real-time feature engineering pipeline using Kafka and Spark Streaming
• Optimized model serving with Redis caching to achieve 50ms average response time
• Deployed using containerized microservices architecture on AWS ECS

Engineered a system to extract insights from unstructured financial PDFs using transformer-based LLMs fine-tuned on domain-specific data.

• Utilized OCR and PDF parsing to extract textual content from scanned financial statements
• Leveraged fine-tuned BERT and GPT-4 models to summarize and answer questions on earnings reports
• Built an RAG (Retrieval-Augmented Generation) pipeline with Pinecone vector search and LangChain
• Deployed system with fastAPI and integrated GPT-based feedback loops to improve accuracy

Developed an advanced analysis pipeline for gravitational wave detection data, implementing machine learning algorithms to identify potential signals in noisy detector data.

• Designed specialized neural network architectures for detecting faint gravitational wave signals in LIGO data
• Implemented time-frequency analysis techniques to characterize signal properties
• Achieved 94% accuracy in detecting simulated gravitational wave events
• Created interactive visualizations to explore detection results and waveform characteristics

Created an interactive web-based dashboard for exploring and analyzing anime ratings, popularity trends, and genre preferences across different demographics.

• Scraped and processed data from MyAnimeList using Python, creating a dataset of 10,000+ anime titles
• Developed interactive visualizations for exploring trends in genre popularity over time
• Implemented recommendation features based on user preferences and viewing history
• Deployed the application using Streamlit Cloud for public access