Final-year Computer Science Engineering student at MIT Academy of Engineering, Pune (CGPA 7.52 · Sem-6 SGPA 8.0), focused on AI/ML, Data Science and Generative AI. I build intelligent, data-driven systems — from agentic RAG pipelines and explainable ML models to end-to-end analytics that turn messy data into decisions.
National Finalist at Insightfy 6.0, IIM Lucknow — out-executed 5,000+ teams to reach the Grand Finale, ranking 1st nationally in Data Analysis and 2nd nationally in the Case Study round. I enjoy turning open-ended problems into shipped, explainable products.
🎯 Core Focus
- 🤖 GenAI & Agents — RAG pipelines, LangGraph state machines, multi-agent orchestration
- 🧠 Machine Learning — gradient boosting, computer vision (YOLOv8, ViT), explainability (SHAP)
- 📊 Data Science & Analytics — EDA, statistical modeling, forecasting, data storytelling
- ⚡ ML Engineering — FastAPI services, vector DBs (FAISS), model deployment
- 🌐 Full-stack delivery — React / Next.js frontends for AI products
|
GenAI & Agents genai = {
"orchestration": ["LangGraph", "LangChain"],
"retrieval": ["RAG", "FAISS Vector DB"],
"agents": ["Multi-agent workflows"],
"models": ["Hugging Face", "LLM APIs"],
} |
Machine Learning & CV ml = {
"boosting": ["XGBoost", "LightGBM",
"Random Forest"],
"vision": ["YOLOv8", "ViT", "GPT-2"],
"explain": ["SHAP", "SMOTE"],
"core": ["Scikit-Learn", "PyTorch"],
} |
|
Data Science & Analytics data_science = {
"wrangling": ["Pandas", "NumPy"],
"analysis": ["EDA", "Statistical modeling",
"Time-series forecasting"],
"viz": ["Matplotlib", "Plotly", "Power BI"],
} |
Engineering & Tools languages: [Python, SQL, Java, C++]
backend: [FastAPI, Flask, REST APIs]
databases: [MongoDB, MySQL, SQLite]
tools: [Git, Docker, AWS, Jupyter] |
An AI-powered financial-intelligence platform driven by 9 specialized asynchronous AI agents coordinating technical analysis, fundamental scanning, and real-time market-regime detection. Built an async multi-key API client with automated failover inside a FastAPI backend for high-availability streaming, and a context-aware News RAG agent using FAISS similarity search to eliminate hallucinations during synthesis.
Stack: Python · FastAPI · React · RAG · FAISS · LLM APIs · yfinance · Vercel
|
Computer-vision credit-underwriting engine — YOLOv8-nano approximates store inventory from field photos, fused with geo-demographic signals and a 10-check fraud matrix for instant alternative credit scoring, explained via SHAP.
|
Autonomous job-matching agent aggregating postings from 7 parallel scrapers, ranking by TF-IDF cosine similarity, and pushing results via async Telegram/email workers.
|
|
Multi-turn conversational state-machine on a LangGraph relational graph — deep-state memory checkpoints drive response drift down by ~90%; intent routing + automated webhooks.
|
Upgrade-propensity classification pipeline — SMOTE within stratified folds, benchmarked ensembles hitting AUC-ROC 0.7969, SHAP-driven revenue buckets (~$33K optimization).
|
|
Multi-modal vision-to-language pipeline — a ViT encoder coupled with a GPT-2 decoder via Hugging Face's VisionEncoderDecoder, with custom PyTorch tensor preprocessing.
|
Multi-city spatial-temporal pipeline — gradient-boosted PM2.5 regressors plus Apriori association-rule mining, delivering 22 automated visualizations.
|
| Achievement | Level | Highlight |
|---|---|---|
| 🥇 Insightfy 6.0 — IIM Lucknow | National Finalist | Beat 5,000+ teams · Rank 1 Data Analysis · Rank 2 Case Study (AUC-ROC 0.7969) |
| Company | Role | Focus |
|---|---|---|
| CodSoft | AI / ML Engineering Intern | Built a multi-modal ViT + GPT-2 vision-to-text pipeline, served via Gradio |
| BB Advisory | Finance & Market Analytics Intern | Technical analysis on FX pairs (USDJPY) & crypto (BTCUSD); risk-management fundamentals |
|
🤖 Anthropic — Advanced Model & LLM Engineering
|
☁️ AWS Academy — Infrastructure & Cloud
📊 Forage — Job Simulations
|
class Rushikesh2026:
currently_mastering = {
"genai": ["Agentic architecture", "Advanced RAG", "LLM evaluation"],
"ml": ["Deep learning", "Model deployment", "MLOps fundamentals"],
"data": ["Advanced SQL", "Feature engineering", "Forecasting"],
}
eager_to_learn = {
"engineering": ["System design", "Scalable APIs", "Vector DBs"],
"cloud": ["AWS in production", "CI/CD", "Docker orchestration"],
"advanced": ["LLM fine-tuning", "Guardrails and evaluation"],
}
goals = [
"Ship production-grade AI/ML systems end-to-end",
"Contribute to open-source AI/ML tooling",
"Deepen data-science and analytics craft",
"Land an AI Engineering / Data Science role",
]
"Learning by building — one real-world project at a time. Build thoughtfully, optimize for tomorrow."
When not coding: reading AI papers · analyzing datasets for fun · competing in case competitions
Fun fact: I spend more time making models explainable than making them accurate — a prediction nobody trusts is just a guess.
