HexaCorp

Data Analyst
May 2025 – Present

• Architected a Retrieval-Augmented Generation (RAG) financial intelligence system using LLMs (OpenAI), FAISS, and Sentence Transformers to extract insights from earnings transcripts, SEC filings, and analyst reports—reducing investment research time by 65%.

• Integrated LangChain-based agent pipelines with real-time financial news feeds, improving portfolio analysts' decision-making accuracy by 22%.

• Designed interactive, retrieval-aware LLM dashboards for revenue forecasting and margin diagnostics across 15+ retail segments, replacing static reporting with dynamic intelligence tools.

• Extracted financial insights from over 9M transaction records using Python (Pandas, Seaborn) and SQL aggregation to support strategic financial planning.

• Developed GPT-powered portfolio commentary tools fine-tuned on historical performance reports to auto-generate quarterly summaries with 90% factual accuracy.

• Automated daily financial and operational reporting via Databricks, reducing reporting latency by 50% and reinforcing production-grade data pipelines.

University of Houston

Researcher – Generative AI Data Scientist
Dec 2024 – May 2025 | Houston, TX

• Deployed biomedical literature search systems using RAG pipelines, prompt engineering, and Hugging Face LLMs—reducing query retrieval time by 70%.

• Built LLM-based QA systems with FAISS and Hugging Face Transformers, improving answer relevance by 30%.

• Automated embedding pipelines using SentenceTransformers and evaluated performance using HELM metrics, enhancing LLM response quality by 28%.

• Streamlined ML deployment workflows with MLflow on Databricks, reducing experiment-to-deployment time by 40%.

• Prototyped multi-turn LLM agents integrated with real-time PubMed queries using Azure OpenAI and LangChain, achieving 82% accuracy.

• Containerized and deployed LLM API endpoints using Docker and Azure Functions, reducing cold-start latency by 40%.

Tata Consultancy Services (Client: AIG)

Data Analyst – Financial Transactions, Claims Forecasting, Payment Risk Analytics
Jan 2021 – Aug 2023 | Chennai, India

• Developed a claims forecasting system using disbursement patterns and seasonal indexing, achieving 92% forecast accuracy for optimized capital reserves.

• Automated ETL pipelines using Event Hub and Apache Airflow to process over 3M daily insurance transactions, cutting latency by 55%.

• Implemented SQL-driven revenue leakage detection across 50+ systems, identifying over $600K in overpayments and anomalies.

• Integrated REST APIs with claims portals for streaming daily financial summaries, reducing manual reconciliation efforts by 60%.

• Built SpaCy + Scikit-learn classification models to triage unstructured claims data, increasing processing speed by 50%.

• Created Power BI dashboards for fraud detection, financial exposure, and claim approval velocity, enabling executive-level visibility.

• Performed variance and sensitivity analysis on 9M+ transactions to uncover operational inefficiencies and guide inventory planning.