I'm most interested in the point where data, AI, and business problems meet. My habit is to understand the actual problem first, work out what the data can genuinely tell us, then build or evaluate the technical solution, and finally explain what changed because of it. The last step matters as much as the rest.
I'm pursuing a Master of Science in Business Analytics at Boston University's Questrom School of Business. Before that I studied Electronics and Communication Engineering with minors in Mathematics and Computer Science, then worked in supply chain analytics and in data analytics and reporting, covering business intelligence and automation along the way. My work has moved steadily toward data engineering, AI and generative AI, analytics, and AI-enabled products.
The engineering background is still the part I use most. It is what makes me ask whether a model is actually measuring the thing we care about, and whether the pipeline behind it will still be correct next month.
Technology gets interesting to me at the moment it improves a decision, a process, or a product.
|
|
The AI technologies above reflect project work and active development rather than claimed mastery.
| Domain | Experience | Focus |
|---|---|---|
| Data Analytics | Professional + Academic | Data exploration, visualization, business analysis and decision support |
| Business Intelligence | Professional | Power BI, reporting, dashboards, DAX, Power Query and stakeholder reporting |
| Data Engineering | Professional + Certification | ETL, pipelines, transformation, data warehousing and Databricks |
| Generative AI | Projects + Current Exploration | LLMs, RAG, AI agents and enterprise AI applications |
| Machine Learning | Academic + Project | Predictive analytics, supervised learning and model evaluation |
| Product Analytics | Projects + Current Exploration | Connecting business and user problems with data and measurable decisions |
| Simulation | Research in Progress | LLM, SLM, statistical and agent-based simulation |
Data Analytics & Reporting Specialist · Aug 2025 – Jul 2026
Analytics, reporting and business intelligence work: using Python and Microsoft Power Platform tools to analyze data, develop reporting, and automate elements of reporting workflows.
Supply Chain Analyst · Nov 2024 – Aug 2025
- Analyzed large-scale datasets to support supply-chain optimization, operational efficiency and cost-focused decision-making.
- Developed data visualizations using Python and Power BI to support strategic decision-making.
- Explored time-series approaches for forecasting supply-chain demand.
|
Questrom School of Business M.S. in Business Analytics August 2026 – August 2027 Boston, Massachusetts Business Analytics · Data Analytics · Machine Learning · Generative AI · Statistical Analysis · Data Storytelling · Business Strategy |
B.S. in Electronics and Communication Engineering 2017 – June 2024 GPA 3.943 · Graduated with Highest Honors Minors in Mathematics and Computer Science A quantitative engineering foundation, then the move into analytics, data and AI. |
Bid Buddy: AI-Powered RFP Intelligence · Enterprise AI × RAG × Product Engineering
Turning a company's scattered proposal history into something you can actually ask questions of while a new RFP is on the clock.
| Problem | Organizations responding to RFPs hold useful knowledge across old proposals, project summaries and historical documents, which makes relevant information hard to retrieve quickly during proposal development. |
| Approach | Upload an RFP, analyze it, retrieve relevant historical company knowledge, and use generative AI to assist with proposal development. Built as part of a Deloitte Innovation Hub initiative, where I worked on the AI stream. |
| Data / Architecture | React SPA into a Java Spring Boot API, with a Python AI worker behind it. PostgreSQL with pgvector for vector retrieval, Google Cloud Storage for documents, Docker on Cloud Run behind API Gateway, and Gemini for generation. |
| Technologies | React · Java Spring Boot · Python · PostgreSQL · pgvector · Google Cloud Storage · Docker · Cloud Run · API Gateway · Gemini · embeddings · vector retrieval · RAG |
| Status | Project work completed within the initiative. |
| Business Value | Shortens the distance between "we have answered something like this before" and having the relevant material and a draft in front of you, with generated text grounded in the company's own documents. |
Capabilities covered RFP upload, document ingestion, chunking, embeddings, semantic retrieval, historical project retrieval, AI-generated proposal sections, and clause and risk analysis.
The part I found most interesting was the knowledge model rather than the model itself. The system stored two record types, DOCUMENT_CHUNK and PROJECT_SUMMARY, which separated granular document evidence from higher-level historical project knowledge. Those answer genuinely different questions: what exactly did we write, versus what have we done that resembles this. A single flat index handles that badly.
Running on Empty: Predicting Bluebikes Bike & Dock Shortages · Urban Mobility × Predictive Analytics × Operations · In Progress
A bike-share station can fail a rider in two ways: no bike available when they want to leave, and no open dock when they arrive.
| Problem | Those two failure modes are the moments the system lets someone down. The research question is: when and where does Bluebikes run out of bikes or open docks, and can those shortages be predicted? |
| Approach | Model shortage risk by station and time window using station identity and capacity, time of day, day of week, incoming and outgoing trip flow, historical demand, real-time availability and weather. |
| Data / Architecture | Bluebikes historical trip data, Bluebikes station information, Bluebikes GBFS live availability, and NOAA historical weather. |
| Technologies | Python · pandas · predictive analytics · public data feeds · time-based feature engineering · supervised learning approaches |
| Status | In Progress. Data assembled and feature work underway; no model results to report yet. |
| Business Value | Identifying stations and time periods with elevated shortage risk informs when rebalancing bikes between stations would be most useful. |
What keeps this one interesting is that the analytical question and the operational question are the same question. A shortage prediction is only worth having if it arrives early enough for someone to move bikes.
Evaluating AI-Driven Simulation in Financial Systems · Research in Progress
If a language model can simulate financial behavior, how would we know whether the simulation is realistic enough to trust?
| Problem | AI-generated simulations are easy to produce and hard to validate. The central question: how can we determine whether AI-generated simulations behave realistically enough to be useful for financial analysis and decision-making? |
| Approach | Evaluate simulation approaches within a financial context and define how to measure sufficient realism, comparing large language models, small language models, statistical simulation, agent-based simulation and hybrid approaches. |
| Data / Architecture | Being scoped. The financial domain and dataset are not finalized. |
| Technologies | LLMs · SLMs · statistical simulation · agent-based modeling · evaluation frameworks |
| Status | Research in Progress. Emerging thesis direction; no experiments run and no results to report. |
| Business Value | A defensible basis for deciding when AI-driven simulation can complement or stand in for traditional methods in financial analysis. |
Evaluation concepts under consideration include statistical similarity, behavioral fidelity, realism, benchmark performance, stability, reproducibility and domain-specific evaluation. At this stage the hard part is not running a simulation, it is deciding what "realistic enough" should mean before anything is measured.
Microsoft
IBM
UiPath
DataCamp
- B.S. in Electronics and Communication Engineering, graduated with Highest Honors
- Undergraduate GPA: 3.943
- Minors in Mathematics and Computer Science alongside the engineering degree
Interested in the space where data becomes a decision, AI becomes a product, and technology has to prove it was actually worth building.