Dakshin JV

• Financial Insights Copilot (RAG)
  Built a Llama3.1-powered assistant with LangChain and MongoDB/SSMS, automating investment recommendations to cut client decision time by 30% and boost platform adoption by 22%.
• LLM-Driven Support Automation Platform
  Engineered a GPT-4 AI ticket routing system with human-in-the-loop refinement, reducing resolution time by 40%. Integrated ZohoDesk APIs, Airflow ETL, and Tableau for real-time SLA tracking.
• AI-Driven Asset Allocation Optimizer
  Engineered a Random Forest model to automate equity/debt allocation by filtering historical anomalies (e.g., Covid-19 outliers), achieving 77% prediction accuracy. The solution cut manual effort by 60%, though it required quarterly retraining to adapt to market regime shifts.
• Automated Mutual Fund Data Aggregator
  Developed a Selenium-based web scraper (Python, BeautifulSoup) to extract historical mutual fund data, enabling strategy refinements that improved client portfolio performance by 15%.
• Life-Milestone Investment Planner
  Built a wealth engine aligning portfolios with client goals (retirement, education), accounting for inflation and volatility, driving 8% new client acquisition in 2 weeks and securing Series A funding.
• Portfolio Rebalancing Engine
  Created a rules-based engine dynamically adjusting portfolios by risk tolerance, yielding 20% higher profits in bull markets, 25% better loss protection in downturns, and 5% new user growth.

• End-to-End Ethereum Data Pipeline
  Built an ETL pipeline decoding blockchain data, cutting costs by 65% using Python, PySpark, and Databricks.
• Blockchain Data Integration
  Aggregated 10+ blockchain sources into a data warehouse for analytics & reporting via S3 and Redshift.
• Social Media Sentiment Analysis
  Analyzed Twitter, Discord, Telegram & Reddit sentiment to forecast token performance using NLP models.
• DeFi Protocol Analytics
  Developed TVL & liquidation metrics for Aave, Compound & MakerDAO by studying whitepapers & on-chain data.
• Credit Scoring System (POC)
  Enhanced creditworthiness predictions by modeling user behavior with ML in Python & PostgreSQL.

GPA: 96%

GPA: 70%