CS student at Scaler School of Technology. I like building things end to end: LLM-powered apps, backend systems that have to stay fast under load, and the CI/CD pipelines that ship them.
- 🔭 Currently exploring RAG, agent environments and DevOps (Docker, Kubernetes, GitHub Actions)
- 🌱 Getting deeper into system design and low-level design in Java
- 🤝 Open to internships and interesting collaborations
- 📫 Reach me at tejasvarshney05@gmail.com or on LinkedIn
| Project | What it does | Stack |
|---|---|---|
| AI Research Paper Synthesis | Turns a pile of papers into grounded, citation-traceable answers and research briefs. arXiv + Semantic Scholar discovery, PDF ingestion, BM25 RAG, BibTeX/RIS export. | React · Node · Express · LLMs |
| Clinical Trial Patient Matcher | OpenEnv RL environment where an agent matches patients from unstructured EHR notes to clinical trials, with tasks that get harder (proxy inference, missing labs). | Python · FastAPI · Docker |
| Search Typeahead | Typeahead over 92k queries: top-K trie, distributed LRU cache on a consistent-hash ring, recency-aware trending, batched writes. ~1–3 ms warm latency. | Python · FastAPI · SQLite · Docker |
| DevSecOps Pipeline | Flask app with a full CI/CD pipeline: tests → CodeQL SAST → pip-audit → Docker build → Trivy scan → GHCR → Kubernetes deploy. | Flask · GitHub Actions · K8s |
| Persona.AI | Chatbot with three distinct personas built purely through system-prompt design. Live demo | React · Express · OpenRouter |