Building Bifrost, the fastest and most scalable AI gateway 🚀

Took Bifrost from concept to production within 3 months, establishing the core architecture and foundations, and now working with a growing team to scale it further.

• 5K+ RPS with <15µs internal overhead on t3.large

• Multi-provider routing with retries & fallbacks

• Plugin-based architecture (plug-and-play extensions)

• MCP gateway (with agentic flow and code mode support)

• Adaptive load balancing

• Fine grained governance

Hiring Go engineers!

Built and scaled Interact, a student-centric networking and collaboration platform used across the university.

• 6,000+ active users for project discovery, collaboration, and opportunity matching

• Scaled to handle 100K+ requests over 7 days with high efficiency

• End-to-end club & event management (tasks, workshops, hackathons, resources)

• Official hosting platform for 50+ events with 5,000+ participants, including VIT Yantra’25

• In-platform meetings & workshops

• ML-powered recommendations, moderation, automation and code reviews

• Microservices-based architecture deployed on GCP with CI/CD

Designed and built the platform end-to-end, owning its architecture, scaling, and production rollout.

Worked on three research papers focused on NLP, Generative AI, and Explainable AI.

• Developed a lightweight NLP model using transformers to analyze climate change policies and assess alignment with Sustainable Development Goals (SDGs), generating data-driven ontologies for better decision-making.

• Enhanced a generative AI model for automated 3D CAD model creation in bioprinting, utilizing medical imaging data (MRI/CT) to generate 3D point clouds with conditional input pipelines.

• Investigated explainability in generative AI models, specifically focusing on latent space in diffusion models to improve interpretability and transparency.

Also contributed to various other researches across the domains in Machine Learning, Computer Vision, and Electronics in limited capacities.

Led and managed all technical aspects of CSI-VIT.

• As a Backend Developer Intern at Fam, I majorly worked on reconciliation and ledger services.

• Developed robust and maintainable automation scripts for streamlining various backend processes.

• Implemented scripts for testing and evaluating queries.