Building AI systems that actually work — LLM routing, RAG pipelines, cloud infra, and winning hackathons at 2am.
Lightweight modular AI routing engine — first of its kind. Routes queries to domain-specialised LLMs (medical, legal, math, code) using a 1.5B router model with VRAM-aware management for consumer GPUs.
88% routing accuracy · 8–20s latency · 6GB GPU support
AI credit appraisal engine built on Azure. Replaced a fragmented multi-step process with a single optimised pipeline using Document Intelligence and GPT-4o for financial document extraction.
99.99% faster · 3 weeks → 3 minutes
AI-powered student productivity platform with a RAG chatbot for institutional search, real-time alerts, automated reminders, and role-based access for admins, CRs, and students.
Decentralised AI training data marketplace. Contributors earn token rewards for quality datasets; AI agencies access verified, curated data with reputation scoring and quality verification.