Work Experience

Data Science Intern at Appsoleut Games

Worked on player retention analytics and churn prediction for a driving simulator game.

  • Analyzed 10K+ gameplay sessions.
  • Built churn prediction models achieving AUC greater than 0.80.
  • Segmented players into low, medium and high-risk cohorts.
  • Provided retention insights for onboarding optimization.

Featured Projects

LLM Gateway — Secure, Cost-Aware Routing Proxy for LLM APIs

  • FastAPI
  • Redis
  • FAISS
  • Docker
  • AWS EC2
  • Prometheus
  • Grafana
  • pytest

Production-grade LLM gateway with circuit-breaker routing, dual-layer caching, prompt injection defense, and per-key budget enforcement — deployed to AWS EC2.

  • Built an OpenAI-compatible gateway with circuit-breaker routing and automatic failover.
  • Blocked adversarial prompts via embedding-similarity injection detection.
  • Added dual-layer caching — Redis exact-match plus FAISS semantic match.
  • Enforced per-API-key budget caps; deployed via Docker Compose to AWS EC2.

ResearchMind — Multi-Agent AI Research System

  • FastAPI
  • LangChain
  • Groq API
  • JWT
  • Prometheus
  • Docker

Production-grade multi-agent research assistant using FastAPI, LangChain, Groq API and agentic AI workflows.

  • Built a 4-step agentic workflow: Search → Scrape → Write → Critique.
  • Implemented JWT authentication and modular FastAPI backend.
  • Added Critic Chain for self-evaluation and hallucination reduction.
  • Integrated Prometheus monitoring and Docker deployment.

YTLens — RAG-Powered YouTube Q&A System

  • FastAPI
  • FAISS
  • HuggingFace
  • RAG
  • JWT

Question answering system over YouTube transcripts using Retrieval-Augmented Generation.

  • Built transcript ingestion and embedding pipeline.
  • Used FAISS vector database with HuggingFace embeddings.
  • Implemented MMR retrieval strategy.
  • Added authentication and persistent storage.
Other Projects 4

Car Price Prediction API

  • FastAPI
  • Redis
  • JWT
  • Prometheus
  • Grafana

Production-grade ML API for real-time vehicle price prediction.

  • Built a FastAPI inference service serving real-time price predictions.
  • Integrated Redis caching to eliminate redundant inference calls.
  • Enforced JWT + API key auth with Prometheus/Grafana observability.

Movie Recommender System

  • Scikit-learn
  • TF-IDF
  • Streamlit
  • TMDB API

Content-based movie recommender using metadata similarity, deployed as a Streamlit app.

  • Engineered tags from overview, genres, cast, and crew.
  • Vectorized text with TF-IDF and computed cosine similarity.
  • Fetched real-time posters via the TMDB API in a Streamlit UI.

SMS Spam Classifier

  • NLTK
  • Scikit-learn
  • Streamlit

NLP-based classifier distinguishing spam from ham SMS messages in real time.

  • Engineered text and structural features (URLs, punctuation, uppercase ratio).
  • Compared Naive Bayes variants; selected BernoulliNB for best precision.
  • Achieved ~97% accuracy with a live Streamlit prediction interface.

Credit Card Customer Churn Prediction (ANN)

  • TensorFlow/Keras
  • Scikit-learn
  • Pandas

Deep learning ANN model predicting bank customer churn from demographic and account data.

  • Preprocessed data with one-hot encoding and StandardScaler.
  • Built a 2-hidden-layer ANN with ReLU and Sigmoid output.
  • Achieved ~86% test accuracy with early stopping to prevent overfitting.

Volunteer Experience

Educator at Nanua Samagra Sewa Samiti

Taught mathematics and science to underprivileged children.

  • Conducted 6+ teaching sessions.
  • Helped students improve academic understanding.

Mentor at Desh Ke Mentor

Mentored students on academics and career planning.

  • Guided 4 students through one-on-one mentoring.