👋 Software Engineer & AI Researcher with a systems-first approach to secure, scalable ML-backed applications.
I am a Computer Science and AI graduate from IIIT-Delhi and currently a Research Assistant at the Network and Systems Security (NetSec) Lab.
My interests lie at the intersection of software engineering, AI systems, and security, with a particular focus on building reliable, failure-aware infrastructure for LLM-powered applications.
I am especially interested in AI safety from a systems perspective, where reliability, policy enforcement, and observability are treated as engineering problems rather than prompt-only solutions.
I am actively seeking full-time software engineering or applied ML roles, where I can contribute to building robust backend and AI-driven systems while continuing to grow as an engineer.
-
AI Governance & Safety Systems
Designing and prototyping fail-closed safety architectures for LLM-based applications, with deterministic pre- and post-inference enforcement and clear auditability. -
Production ML & MLOps
Building and evaluating ML pipelines using Google Cloud (Vertex AI) and BigQuery, with experience in parameter-efficient fine-tuning (LoRA) and applied ML for security use cases. -
Backend & Distributed Systems
Developing and improving backend services using modern web stacks, including work on high-traffic payment systems and low-level Unix-based environments. -
Applied Computer Vision
Working on real-time vision-language systems (YOLO, LangChain) for assistive and safety-critical applications.
AI Governance Gateway (Research & Systems Engineering Project)
A reference architecture exploring deterministic safety enforcement outside the LLM execution engine.
- Goal: Explore how LLM safety can be enforced at the infrastructure layer rather than relying solely on model behavior.
- Design: Dual-gate (pre/post-inference) enforcement, Zero-Trust–inspired boundaries, and immutable audit logging.
- Tech Stack: FastAPI, n8n, Supabase (PostgreSQL), Microsoft Presidio, Gemini.
- View Repository
- Read Security & Design Disclosure
| Area | Project | Key Outcome |
|---|---|---|
| AI Safety | Project Aether | Designed a systems-level approach to LLM governance and auditability. |
| Security | Side-Channel Malware Detection | Achieved 97% Macro F1 using syscall-based behavioral analysis. |
| FinTech | Payments Revamp (SDE Intern) | Reduced transaction failures by 25% and improved latency by 30%. |
| ML Optimization | LLM Fine-tuning (LoRA) | Reduced average inference time by 40% for domain-specific assistants. |
| NLP | Emotion-Cause Recognition | Achieved 99.4% F1 on structured emotion classification tasks. |
- Languages: C, C++, Java, JavaScript, Python, SQL, TypeScript
- AI / ML: PyTorch, TensorFlow, HuggingFace, BERT, Scikit-learn, LangChain
- Backend & Infra: FastAPI, Docker, Kubernetes, AWS, Google Cloud Platform (GCP)
- Data Systems: PostgreSQL, MySQL, MongoDB, BigQuery
“Architecture is not about what it is — it’s about what it prevents.”