I make data do the work. Lately that’s a drone that delivers packages on its own, a chatbot that turns questions into SQL, and a model that talks like me.
Who I am. Curious about how things work and how to make them better. I sit where data, AI and software engineering meet, and I care about simplicity and good design.
What I build. Data systems that make complex information useful: pipelines, cloud analytics, business intelligence and AI applications. I like owning the whole lifecycle, from the problem and the architecture to the last screen a user sees, and the engagement analytics beyond it.
How I work. Understand the problem before solving it. I ask the right questions, check the constraints and weigh the trade-offs, then keep refining until the work is fast, reliable and simple.
How to reach me. Email me, or write me a postcard at the bottom of this page. My résumé is at the top; LinkedIn and GitHub are at the bottom.Email me
Selected work
- Unmanned aerial vehicleAutonomous systemsAn autonomous drone that delivers packages across a city. Patented in India.
An autonomous drone that delivers packages across a city. Patented in India. View project
- Movie chatbot for structured dataLLM agentsFour Bedrock agents turn plain-English questions into safe, audited SQL. 98% answer accuracy.
Four Bedrock agents turn plain-English questions into safe, audited SQL. 98% answer accuracy. View project
- Talk LikeLLM fine-tuningLlama 3 fine-tuned on nine years of my chats, with memory. 94.88% accuracy on conversational nuance.
Llama 3 fine-tuned on nine years of my chats, with memory. 94.88% accuracy on conversational nuance. View project
- Lung segmentation in CT scansMedical imagingNeural networks that segment the lungs automatically. 87% accuracy.
Neural networks that segment the lungs automatically. 87% accuracy. View project
- Image caption generatorComputer visionCNN and LSTM captions that beat Google NIC on ROUGE and METEOR.
CNN and LSTM captions that beat Google NIC on ROUGE and METEOR. View project
- Multi-evidence natural language inferenceClinical NLPLlama 2 and PubMedBERT on breast cancer trial reports. 66.74% F1.
Llama 2 and PubMedBERT on breast cancer trial reports. 66.74% F1. View project
- Spotify music analyticsAnalytics95,000+ tracks, hypothesis tests and regression for recommendations.
95,000+ tracks, hypothesis tests and regression for recommendations. View project
Watch the movie chatbot work
An animated walkthrough, not the live chatbot. Pick a question to replay how four Bedrock agents clarify what you mean, write the SQL, wait for your confirmation, run it safely, and explain the answer in plain English.
- Orchestrator (done)
- SQL Generator (done)
- Execution (done)
- Explainer (done)
Replay with sample data. Nothing here is sent anywhere.
Which genres made the most at the box office since 2000?
Do you want total box office for each genre, or the average per film?
- Validating the SQL: read-only, movies schema only
- Checking the cost with EXPLAIN
- Running on Redshift Serverless through the Data API
- Writing the audit record to S3
Done. Cancelled automatically if a query ever runs past 110 seconds.
WITH genre_revenue AS (
SELECT g.genre_name,
SUM(m.box_office_revenue) AS revenue,
COUNT(*) AS films
FROM movies.movies m
JOIN movies.movie_genres g ON g.movie_id = m.movie_id
WHERE m.release_year >= 2000
AND m.box_office_revenue IS NOT NULL
GROUP BY g.genre_name
),
ranked AS (
SELECT genre_name, revenue, films,
RANK() OVER (ORDER BY revenue DESC) AS rnk
FROM genre_revenue
WHERE films >= 50
)
SELECT genre_name, revenue, films
FROM ranked
WHERE rnk <= 3
ORDER BY rnk;$41.2B
Action leads on total box office since 2000, ahead of Adventure and Comedy, which release more films but earn less each.
| Genre | Total box office | Films |
|---|---|---|
| Action | $41.2B | 612 |
| Adventure | $37.9B | 448 |
| Comedy | $29.4B | 901 |
Debug: statement_id 3f2a…c91e · exec_time_ms 1,284 · row_count 3
98% answer accuracy100% SQL execution successEvery query confirmed by a human
A visual representation with sample data, not connected to the live chatbot. How it’s built
A drone that crosses a city on its own: it flies, finds its way, and drops the package.
Patented in India as “Unmanned Aerial Vehicle (UAV) and Navigation Method Thereof.” How it worksSkills
Five things I do well. Each one opens to the proof.
About me
- Based in
- Seattle, WA
- Currently
- Data Analyst at Amazon
- Studied
- MS in Data Science, CU Boulder
- Also known as
- Scooby
I studied electronics and telecommunication at PICT in Pune, with honors in data science, and built the drone models that became a patent. My master’s in data science at CU Boulder took me into NLP and LLMs. Now I build analytics and AI tools at Amazon, and on my own time, things like a model that talks like me.
On record
Patent and papers
- Unmanned Aerial Vehicle (UAV) and Navigation Method ThereofPatent, 2022
- Cascaded PID Control System for UAV with Gain Factor Prediction Using MLIJRASET, 2021
- Path Planning and Obstacle Avoidance for UAV in an Uncharted EnvironmentIJRASET, 2021
Education
- Master’s in Data Science, University of Colorado Boulder2024
- BE in Electronics and Telecommunication, Honors in Data Science, PICT2022
Off the clock
In front of the camera
I modelled from 2017 to 2022, which mostly meant learning to look thoughtful on command. It paid off: I ended up on Levi’s billboards.
Click the front photo to read the back.
