CAMS Lens — SEBI-Grounded Compliance Intelligence Platform
Productionized an enterprise-grade GenAI compliance platform leveraging RAG over 80,000+ SEBI and financial regulatory documents. Achieved 95% retrieval precision with sub-7-second response latency, delivering audit-ready, citation-backed outputs in regulated financial environments. The system auto-ingests new files without human intervention and scales automatically with growing regulatory updates.
- Role: AI Architect & Lead Engineer (Solo Developer)
- Timeline: June 2024 - Present
- Team: Independently built core AI, backend, and deployment infrastructure
- Technologies: RAG, Google Cloud, Vertex AI, Cloud Run, AlloyDB, BigQuery, Python, AI/ML
- Link: https://camslens.camsai.com/
Problem
Mutual Fund compliance and research teams manually reviewed 80,000+ SEBI circulars, AMFI guidelines, and regulatory documents, spending 4–7 hours per analyst daily. The process was fragmented, interpretation-heavy, audit-sensitive, and failed to scale with increasing regulatory updates.
Solution
Deployed a SEBI-grounded Retrieval-Augmented Generation (RAG) system enabling compliance officers to query regulatory documents in natural language and receive precise, citation-backed answers. Built structured regulatory taxonomy (27 intermediaries, 150+ subcategories), persona-based outputs (Compliance, BRD, SOP, FAQ), and automated RSS ingestion for real-time circular updates.
Impact
- 95% retrieval precision with citation enforcement
- Reduced regulatory research time from 4+ hours to under 5 minutes
- Indexed 80,000+ regulatory documents with structured metadata
- Processing 1,000+ compliance queries daily in production
- Enabled automated BRD and SOP generation for regulatory changes
- Publicly covered by major financial publications
Key features
- SEBI-grounded RAG engine with strict citation validation
- Custom retrieval ranking beyond cosine similarity (metadata-aware scoring)
- Persona-based outputs: Compliance checklist, BRD, SOP, FAQ
- Regulatory taxonomy across 27 intermediaries & 150+ subcategories
- Automated RSS ingestion pipeline for real-time circular updates
- Multi-document contextual synthesis with traceable references
- Query analytics and usage monitoring dashboard
- Audit-ready transparency with document and page-level citations
Tech stack
- Frontend: React, TailwindCSS, Chart.js
- Backend: Python, FastAPI
- Ai: Vertex AI (Gemini), Custom RAG Pipeline, ChromaDB
- Cloud: Cloud Run, BigQuery, Pub/Sub, Cloud Storage, AlloyDB
Architecture highlights
- End-to-end ingestion pipeline: RSS → Parsing → Chunking → Embedding → Indexing
- Metadata-aware vector filtering (doc_type, intermediary, category)
- Prompt templating with compliance-focused grounding constraints
- Latency optimization via async retrieval and batched embeddings
- Monitoring: Query logging, citation validation, failure tracking
What K Laxman learned
- Designing production-grade RAG systems in regulated financial environments
- Mitigating hallucination through strict grounding and citation enforcement
- Balancing retrieval precision vs latency under enterprise constraints
- Building scalable document ingestion pipelines with fault tolerance
- Translating regulatory circulars into operational artifacts (BRD/SOP automation)
- Optimizing GenAI cost-performance tradeoffs in production
Press coverage
- CAMS introduces AI tool for real-time regulatory analysis — Angel One
- CAMS’s technology and AI innovations powering growth — Economic Times
Explore more
- Home — overview, skills and a built-in AI assistant
- Experience — roles at Think360 AI (CAMS), CAMS Mutual Funds and IIT Delhi
- Projects — GenAI, LLM, RAG and full-stack builds
- Education — IIT Delhi, M.Tech & B.Tech Computer Science
- GitHub Activity — open-source contributions
- Contact / Hire me