I'm Fardeen, an MS CS student at GMU with a B.Tech background from VIT. I'm driven by the intersection of software engineering and artificial intelligence—building systems that are scalable, intelligent, and user-centric.
What I do: Full-stack development (MERN, FastAPI, Spring Boot) | ML/AI systems (TensorFlow, PyTorch, scikit-learn) | Cloud & DevOps (AWS, GCP, Docker, Kubernetes)
Current focus: Healthcare ML projects, fintech solutions, loan prediction systems, and resume screening assistants. I love writing clean code, deploying production systems, and tackling hard problems.
Open to software engineering and AI/ML internships. Always learning, always building.
| Project | What it is |
|---|---|
| DocAgent | Agentic RAG assistant — hybrid FAISS + BM25 retrieval, cross-encoder rerank, ReAct agent with grounded, cited answers. FastAPI, Redis workers, React UI, Kubernetes, Terraform. 100% recall@5, 92.5% answer accuracy on a 40-question eval. |
| Intent Classifier MLOps | DistilBERT + LoRA on banking77 — 91.5% accuracy vs 88.8% TF-IDF baseline. MLflow tracking, int8 ONNX export (268 → 68 MB, 99% parity), FastAPI serving at ~1.7 ms/text, Docker + CI. |
| Switchyard ML | ONNX model-serving platform with dynamic batching, canary releases, automatic rollback, Prometheus metrics and a React console. |
| Sagacious-AI | Local multi-model workspace — parallel candidates, blind rubric judging, streamed progress and SQLite history. |
| Project | What it is |
|---|---|
| AI SQL Writer | Natural-language → SQL with schema discovery and conversational data exploration (FastAPI + React). |
| Contract & Invoice Intelligence | Invoice PDF/text extraction, contract validation and a review queue. |
| Retail Recommendation Platform | SVD recommendations, RFM/KMeans segmentation and Apriori rules with a FastAPI/React dashboard. |
| Loan Default Modeling | scikit-learn + PyTorch models with SHAP explainability and a Streamlit dashboard. |
| Parkinson's Hybrid Prediction | Random Forest/SVM hybrid on vocal features with subject-disjoint evaluation. |
| GenAI Traffic Fingerprinting | Packet-metadata toolkit — PCAP import, six classifier baselines, early-window evaluation. |
| Project | What it is |
|---|---|
| My J.A.R.V.I.S | Animated voice assistant for macOS with local Ollama conversation. |
| Aleesa Commerce | Next.js + Prisma store with inventory, transactional orders and Razorpay. |
| Student Survey App | Angular + Spring Boot CRUD app with REST, JPA and MySQL. |
| Container Delivery Pipeline | Docker, Jenkins and Kubernetes deployment for a survey site. |