A curated mix of cutting-edge AI/ML experiments and full-stack applications highlighting my diverse skill set.
Python, LangChain, LangGraph, Transformers
- Production-ready candidate vetting assistant with automated question generation and scoring.
- Delivers 85% validation accuracy while shrinking hiring cycles by 60%.
FastAPI, pgvector, FAISS, LangGraph
- Automates 90% of grading pipelines with RAG-based evaluation, plagiarism detection, and question synthesis.
- Reduced evaluation time by 75% while keeping human-in-the-loop controls.
LangChain, FAISS, Hugging Face, FastAPI
- Converts natural language into SQL for 300+ CSV datasets with 92% accuracy.
- Ships lightweight analytics copilots that cut manual reporting effort by 80%.
Python, scikit-learn, XGBoost, SQL
- Hybrid collaborative-filtering recommender serving 250K+ users with optimized ETL pipelines.
- Increased click-through rates by 18% through feature engineering and parallel training.
TensorFlow, FastAPI, React, OpenCV
- AI-powered healthcare companion delivering 87% diagnostic accuracy and conversational guidance.
- Integrates symptom analysis, recommendation engines, and secure user dashboards.
Python, Whisper, Transformers, Streamlit
- Summarizes multi-speaker conferences with diarization, topic mining, and action-item extraction.
- Reduces meeting review time by 75% for distributed teams.
React, FastAPI, PostgreSQL, OpenCV
- Microservices SaaS for auditing AI-generated media with 100MB+ uploads, background jobs, and live telemetry.
- Ensemble inference pipeline surfaces risk scores in under 10 seconds.
Node.js, Express, MongoDB, LangChain, Chrome APIs
- Browser extension that logs reading journeys and generates AI analogies synced across tabs.
- Backed by Groq Cloud inference with <500 ms latency dashboards.
React, Node.js, FastAPI, MongoDB, OpenCV
- Image-processing CMS that supports 10K+ daily uploads with 500 ms request latency.
- Optimized REST APIs, batching, and queueing ensure high concurrency and 92% classification accuracy.
FastAPI, MySQL, Django ORM, Pydantic
- Factory + Observer design patterns deliver 150 ms average lookups for contact intelligence.
- Embedded spam detection raised accuracy by 90% for inbound leads.
