Srikanth K Nattudurai User

I’m a B.Tech Computer Science and Engineering student who builds production-style systems with a focus on scalable, observable behavior—especially where LLM safety and evaluation matter. I also actively contribute to open source, improving performance and interoperability in widely used scientific tooling.

In OpenMS, I merged multiple pull requests and helped extend Python interoperability through modern C++/Python bindings. I implemented zero-copy NumPy structured array bindings via nanobind, enabling up to 30x speedup at 1M peaks, and I extracted a reusable SequenceCoverage API to support downstream Python analysis.

Alongside open source work, I engineered software projects like a Guardrail Gateway and an LLM Eval Pipeline using FastAPI, React, and PostgreSQL. I built an input/output prompt safety gateway, added a large jailbreak/prompt-injection rule set, and strengthened it with an ML injection classifier and adversarial testing (305 automated tests).

I’m drawn to engineering that ties together concurrency, provider abstraction, and measurable quality gates—like async evaluation jobs, persisted evaluation history, and CI regression thresholds (80% pass rate). I want to keep growing in backend systems and AI/LLM engineering through collaborative, impact-driven work.