My Work

Built a full-stack race strategy simulator using Next.js, FastAPI, and Gemini AI, enabling real-time simulation, multi-strategy comparison, and weather-aware race prediction with interactive visualizations; deployed on AWS Lambda and Vercel, achieving 40% faster cold-start time through serverless optimization and AI inference rate-limiting.

A cloud-native microservices platform for real-time business recommendations using gRPC, Kafka, Redis, and Flask. Built scalable AI-first architecture with PostgreSQL, Kubernetes, and content-aware recommendation engine.

Built a job-matching system using React, Django, and GPT-4o, exposing REST APIs for LLM-based resume tailoring and similarity scoring; improved match relevance by 20%.

A machine learning project predicting F1 championship winners using clustering, PCA, and ML models. Uses data from the Ergast F1 API and explores driver and constructor patterns.

Engineered a real-time recommendation system using Kafka, Redis, and FastAPI, with Docker Compose orchestration and async data pipelines, reducing API latency by 35% through smart caching and efficient I/O handling.

Developed a production-grade sentiment analysis API using FastAPI and BERT, with Prometheus-based monitoring and Dockerized deployment; achieved 98% accuracy on IMDb and enabled PyTorch-TensorFlow portability.

A data visualization dashboard that explores Formula 1 racing statistics using Streamlit and embedded Tableau Public views. Provides real-time, interactive insights into race data, driver performance, and team analytics.