Press Enter to continue
Initializing modules...
CURRENT MODULE: —
BUILDING AI SYSTEMSFOR REAL-WORLDCONSTRAINTS
Architecture is not just code. I build AI systems engineered for real-world constraints—bridging complex LLM capabilities with predictable environments.
Core Philosophy
Systems Thinking as a Standard.
Architecture is not just code; it's the invisible connective tissue between human intent and machine execution. I design AI-native platforms where system boundaries and real-world constraints matter.
From orchestrating multi-agent workflows to scaling RAG inference, my focus is bridging complex LLM capabilities with predictable, explainable environments.
Project Repository
Selected Deployments
DocuMind
Production-grade document intelligence connecting unstructured PDFs to LLM contexts. Engineered a resilient async ingestion pipeline delivering context-isolated, SSE-streaming resolutions without stalling the main event loop.
Signal (ACIA)
Autonomous intelligence pipeline scraping dynamic pricing pages. Built a cost-aware delta engine that bypasses expensive LLM inference loops unless statistical pricing changes exceed a 5% baseline discrepancy threshold.
LoanWise AI
An origination workflow orchestrated by sequentially integrated AI agents. Engineered dynamics fusing regulatory heuristics seamlessly to enforce bias-free, deterministic fallback conditions ensuring resilient approvals.
SpaceFlow
Workspace utilization platform neutralizing ghost-booking overhead. Bridged hardware logic with an intelligent recommendation backend relying on deterministic fallbacks to align real-time space optimizations with intent.
VoyageAI
Travel orchestrator structuring natural language into itineraries. Embedded contextual AI telemetry within map boundaries using state-preserving React caching blocks and unified server pipelines.
Leaf Inference Engine
Edge-adjacent computer vision classification backend executing custom CNN operations. Streamlined pipeline isolating deployment logic from a high throughput FastAPI microservice that scrupulously scores 38 pathological states.
Academic Training
Education Foundation
M.Tech — Software Engineering
National University of Singapore
Focused on systems architecture, AI systems, and designing scalable, production-grade software. Evaluated trade-offs in real-world environments bridging complex LLM capabilities with predictable constraints.
M.Tech — Software Engineering
National University of Singapore
Focused on systems architecture, AI systems, and designing scalable, production-grade software. Evaluated trade-offs in real-world environments bridging complex LLM capabilities with predictable constraints.
B.Tech — Information Technology
Vellore Institute of Technology
Built strong foundations in computer science, software engineering, and full-stack development covering core programming concepts, database design, and web technologies.
B.Tech — Information Technology
Vellore Institute of Technology
Built strong foundations in computer science, software engineering, and full-stack development covering core programming concepts, database design, and web technologies.
System Trust
Certifications
Architecture Logs
Documentation & Writing
RAG without the theater
Chunking, search, and citations beat demo polish. Notes on document QA that still works after the meeting ends.
Why we build what we build
Name the problem before the stack debate runs away. How I try to keep intent and architecture pointed at the same thing.
What to build is still the hard part
Lots of noise about tools and prompts. Almost none about which problem deserves your month. Where I am putting attention.
Claude and agentic building
Tight loops with Claude over your repo and terminal. What helped, what lied, what I still own.
Ready to build real systems?
I'm currently exploring early-career roles and internships. I care entirely about building resilient AI systems and backend architectures that actually run in production.