Hey, I'm
$ git log --oneline --graph
Five years of turning raw data into dashboards people actually use set the direction. Every stop since has been about shortening the distance between a number and a decision.
feat: SQL-based validation + reporting framework, -60% discrepancies
First real lesson in production reporting: a 4.5M-customer billing base doesn't forgive a wrong number on a dashboard. Built row-level checks and reconciliation reports that cut discrepancies 60% and shortened month-end close by 10%.
feat: automated KPI dashboards, 99% on-time delivery
Moved from ad hoc reporting to automated KPI dashboards (churn, usage patterns) that leadership could track every week, with 98%+ accuracy holding across 600K+ subscriber billing records. Earned "Best Upcoming Talent," FY23 Q3.
feat: candidate scoring analysis + compliance interface, 96% accuracy
Delivered a resume-to-research-fit scoring analysis (TF-IDF → SBERT → Neural Refinement) for RA screening, and a 5-tab compliance interface visualizing 5 CMS-mapped rule findings for MyRaina's AI fall-monitoring platform.
feat: leadership dashboards shipped to production
Where the two threads met: a six-tab FinOps dashboard surfacing $146K in platform spend, and a 211-check reporting framework keeping 23 production tables clean and reportable.
$ cat PRINCIPLES.md
Everything I build is grounded in a production mindset.
"Understanding existing systems deeply before changing them, designing for the next person who has to maintain my work, and communicating technical findings in a way non-technical stakeholders can act on." That’s the standard I actually build to.
The Cost Analytics dashboard didn't stop at showing total spend. It isolated $13.2K in idle cost across 885 specific unused objects, so the finding pointed straight to an action, not just a KPI.
→ see it across Experience 02The Penn State Alumni event model started as 6 disparate Excel sources with no shared definition of a KPI. Consolidating them into one model was the real work; the forecast was just what came after.
→ see it in Alumni Event Forecasting 03Every forecast I ship comes with a Conservative / Expected / Optimistic scenario, not a single number, because a stakeholder making a budget call needs to see the range they're actually planning against.
→ see it in Alumni Event Forecasting 04The Alzheimer's Market analysis shipped as parallel Python and Tableau dashboards on purpose: the same finding, in the tool each audience already trusts, not a screenshot exported from whichever tool I personally prefer.
→ see it in Alzheimer's Market Overview$ cat experience.log
Barton Malow Featured
Pennsylvania State University
Oracle Corporation
Accenture Solutions Pvt Ltd
$ ls ./projects
TRx-based pharmaceutical market analysis across brands, specialties, and channels. 202,499-row dataset, built as parallel Python (Plotly) and Tableau dashboards for direct platform comparison.
Multi-year (2022–2025) event performance analysis with scenario-based 2026 forecasting, projecting 235 attendees and $49/attendee revenue via a blended trend + recency-weighted model.
Deep Dive · Independent Project, 2026
A 202,499-row TRx pharmaceutical dataset had real signal in it, brand share shifting, specialty channels performing differently, but no single dashboard the whole stakeholder audience actually trusted.
A brand manager who lives in Tableau and an analyst who lives in Python shouldn't have to fight over which tool the truth lives in. The finding matters more than the tool it's rendered in.
The same brand/specialty/channel analysis, built twice: once as an interactive Python (Plotly) dashboard exported as a shareable standalone HTML file, once as a native Tableau workbook, so either audience gets the same answer in the tool they already trust.
This is where principle 04 (build for the audience, not just the tool) isn't an abstraction: the same brand/specialty/channel analysis ships twice, once in Tableau for stakeholders who live there, once in Python for anyone who wants to dig into the code.
Deep Dive · Pennsylvania State University, 2026
Six years of alumni event data lived in 6 disparate Excel files with no shared KPI definitions, so planning for the next event meant starting from scratch and guessing at the budget.
A budget conversation without a range is just a guess dressed up as a number. Stakeholders needed to see best-case and worst-case, not one point estimate to argue about.
Consolidated 6 Excel sources into one unified KPI model, then built a blended trend + recency-weighted forecast with three funding scenarios, conservative, expected, optimistic, so planning moved from reactive to data-informed.
This is where principles 02 (one source of truth beats six spreadsheets) and 03 (show the range, not just the point estimate) aren't abstractions: six Excel files became one model, and the forecast ships as three scenarios, not one guess.
$ cat research.md
Accepted and presented at SEET 2026 (2nd International Conference on Software Engineering of Emerging Technologies), Penn State Behrend, PhD Research Track. Listed on page 76 of the conference abstract book. Now published in the official conference proceedings by Springer Nature (CCIS vol. 3126).
Read on Springer → View conference →Won the Prototype Phase and MVP Phase, advanced through funded phases to a Top 10 finish at the Pitch Contest, Hintz Family Alumni Center, University Park.
Recognized for exceptional contributions and high-impact performance on the Telia Company billing analytics engagement.
Representing students in discussions with university leadership; championing AI literacy and equitable AI access across Penn State's Commonwealth Caucus.
$ cat skills.yaml
Tableau · Power BI · Plotly · Streamlit · Data Visualization · Matplotlib · Seaborn
Python (Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn, TensorFlow, Keras) · SQL · R · Java
Time Series Forecasting · Scenario Modeling · Regression · Reconciliation · KPI Design
Claude Code · Claude CLI · Claude Cowork · Claude Skills · Claude Projects · Claude MCP · GitHub MCP · ChatGPT · GitHub Copilot · Cursor IDE · Perplexity · Ollama · Google Gemini · Julius AI
Databricks · Delta Lake · PySpark · Unity Catalog · Azure · AWS · Oracle Database · PostgreSQL · PL/SQL · Advanced Excel
Git · GitHub Actions · Jupyter · Jira
$ cat education.log
Master of Data Analytics · GPA 3.96/4
Statistical Analysis, Data-Driven Decision Making, Predictive Analytics, Deep Learning, Natural Language Processing.
B.Tech, Electronics & Instrumentation Engineering · GPA 3.3/4
Control Systems, Digital Electronics, OOP, Data Structures & Algorithms, Web Technology, Artificial Intelligence.
$ mail tushar
Open to full-time Data Analyst / Business Intelligence Analyst roles starting Dec 2026.