About Me

I’m a Data Scientist and ML Engineer based in Boston, Massachusetts. My work spans Applied Deep Learning, Data Engineering, and Network Science — with a growing focus on end-to-end ML product development and applied research.

Previously, I worked in Credit & Fraud Risk Data Science at American Express, contributing to Global Collections and US Commercial Limits teams across modeling and risk strategy.

I’m open to exciting opportunities and collaborations — let’s connect!

---

Latest Work

### CineSphere — Movie Recommending Chatbot Using Knowledge Graphs - Developed a chatbot using GPT-4 with multi-agent LangChain Framework and a Knowledge Graph over 45K movies - Leveraged OpenAI Function Calling, Vector Indexing, and Node Similarity algorithms for robust recommendations

[](https://github.com/shu3hamiitkgp/CineSphere)

### LES Weather Forecasting — Multi-Modal Model - Designed an Encoder-Decoder architecture using ConvLSTM and LSTM for weather forecasting at Lake Michigan - Combined satellite imagery with meteorological data; achieved recall of 0.85 on highly imbalanced data

[](https://github.com/shu3hamiitkgp/LES-Precipitation-Forecasting-Multi-Modal-Architecture)

### TravelBud — AI Tool for Travel Planning - Built a travel itinerary-generating app optimized for user interests, budget, and destination - Leveraged RapidAPI for flight & hotel data and OpenAI GPT-3 for chat completion in a cloud ecosystem

[](https://github.com/shu3hamiitkgp/TravelBud)

### A/B Test — Understanding Causal Relationships in Experiments - Published an in-depth article explaining A/B test implementation and how they reveal causal relationships

[](https://medium.com/aiskunks/a-b-test-understanding-causal-relationships-in-experiments-2a7a0661b276)

### DIME Network Analysis — US Federal Election Money - Researched DIME data to map monetary contribution networks in US Federal Elections - Applied network science to reveal scale-free properties and ideology-based contributor communities

[](https://github.com/shu3hamiitkgp/DIME-Network-Analysis)

### GrocerEase — Smart Digitization of Groceries *(Founder)* - Built an app that digitizes grocery receipts via photo capture using Deep Learning and Computer Vision - Automatically tracks items, quantities, and estimates expiry dates to reduce waste - **Top 5 finalist** competing for $10K seed funding at Husky Startup Challenge, Northeastern University

[](https://drive.google.com/file/d/16PXPgmUzvlOt_x3_S7xSzGZnyMC3hiVO/view?usp=drive_link)

### VaniVerse — Conversational AI for Audio - Conversational AI platform that generates prompts from uploaded audio files - Leveraged OpenAI Whisper & GPT API with Airflow for concurrent audio processing and prompt generation

[](https://github.com/shu3hamiitkgp/VaniVerse)

### Hindi Seq2Seq Model — Generating Hindi Poems with RNN - Trained a Sequence Model on a corpus of Hindi poems to generate new text using NLP and text processing - Used pre-trained Hindi tokenization and word embeddings from AI4Bharat

[](https://github.com/shu3hamiitkgp/Hindi-Seq2Seq-Model--Generating-Hindi-Poems-using-RNN)

### Runner's High — Reflecting on Runs in '22 In 2022, I set a goal to run 500 km over the calendar year. I used Python to visualize geospatial data collected via Strava, sharing insights on the journey.

[](https://github.com/shu3hamiitkgp/strava_py) [](https://medium.com/@goyal.shubh15/runners-high-reflecting-on-runs-in-22-723b991094ea)