I'm Afsha, a ๐ค Data Scientist & AI Enthusiast who loves building intelligent systems that solve real problems. Naturally curious and detail-driven, I care deeply about understanding the why behind every model โ not just the output.
On the technical side, I'm deeply invested in Machine Learning, Deep Learning, Generative AI, and Computer Vision. I love working end-to-end โ from raw data wrangling and model training to deploying live applications on cloud platforms.
I recently completed my M.Sc. in Data Science from Aligarh Muslim University โ and I'm just getting started. Every project since has been about pushing what I know further and building things that matter.
End-to-end agentic retrieval-augmented generation pipeline with multi-hop reasoning, tool-use, and dynamic context retrieval โ grounded in knowledge bases for accurate, traceable answers.
Full-stack crypto analysis system combining a Markov-chain predictor, a market-intelligence aggregator (Fear & Greed Index, CoinGecko), and a Kelly Criterion risk manager, with LLM-generated commentary via Groq.
A dark-themed loan calculator with live EMI computation, full amortization schedule, principal-vs-interest visualization, and a prepayment simulator that shows real interest savings.
A six-stage multi-model fusion pipeline using Florence-2-Large, BLIP ITM, and Qwen2.5. Evaluated on Flickr8k benchmark (8,091 images). Deployed on HuggingFace Spaces.
Classified news as real or fake using TF-IDF vectorization across Logistic Regression, SVM, Naive Bayes, and Decision Tree with comparative evaluation and visualization.
Analyzed cryptocurrency trading behavior integrating Hyperliquid data with the Crypto Fear & Greed Index to uncover how market sentiment influences trader profitability.
Parcel-level heat risk analysis of a 75-acre Indianapolis athletic complex using FortyGuard API endpoints, with heat exceedance and persistence metrics. Found 96.9% impervious surface and a 45.6ยฐC peak heat index.
Three-class health risk classification scored on balanced accuracy, using LightGBM tuned with Optuna. Rank 608 / 3,355 teams.
Tabular prediction challenge from Kaggle's Playground Series, Season 6 Episode 8. Rank 1,359 / 3,531 teams.
Dept. of Statistics & Operations Research
Aligarh Muslim University (AMU)
I.P. College (Campus-2)
CCS University