Syed Mohammed Abrar Mohtasim@ABRAR
1
Events attended
1
Submissions made
Bangladesh
1 year of experience
About me
I am an Agentic AI Engineer who builds production-grade AI systems where LLMs go beyond answering questions — they plan, route tasks, call external tools, manage conversational context, verify outputs, and automate real business workflows.My professional background in enterprise data management at Hirdaramani Bangladesh gave me firsthand exposure to data quality problems — which directly motivated my focus on building reliable, grounded AI systems. My strongest work is in: Agentic AI Engineering — designing multi-agent pipelines with role specialization, sequential orchestration, tool use, and query routing LLM Engineering — prompt design, output grounding, anti-hallucination systems, fine-tuning, and evaluation AI Product Engineering — turning agent prototypes into deployable systems with databases, dashboards, APIs, and automation layers I care about building AI that is not just impressive in demos — but reliable, verifiable, and genuinely useful in production.
🤓 Latest Submissions
Chat with Your Enterprise Data - AI Data Analytics
Chat with Your Data is an enterprise analytics platform that transforms how organizations interact with their data. Built for Track 4: Data & Intelligence, it delivers complete coverage of all five focus areas: RAG over multi-source data, natural language querying, knowledge graph extraction, anomaly detection, and AI-powered data pipelines. The system leverages Gemini 2.5 Flash in three powerful ways: native function calling for database queries (eliminating SQL hallucinations), knowledge graph extraction from PDF documents, and Google Search grounding for cross-source question answering. Every query is processed through Veea's LobsterTrap security firewall, which inspects prompts before they reach the AI, blocking prompt injection, PII requests, and data exfiltration attempts with only 50ms overhead. Our core innovation is a deterministic SQL builder that generates perfect queries from natural language, ensuring dashboard numbers, chat answers, and monitoring alerts always match—solving the consistency problem that plagues traditional AI analytics. The system processes 24 KPIs through parallel execution, delivering sub-2-second dashboard loads and 3-8 second chat responses. Five integrated features demonstrate production readiness: an executive dashboard with real-time KPIs, a conversational chat interface, four-layer anomaly detection, document intelligence with interactive knowledge graphs, and security diagnostics with live canary tests. The architecture scales from 13 properties to thousands without modification, handles years of historical data, and maintains enterprise-grade security throughout. Built with Streamlit, Supabase PostgreSQL, and deployed on Hugging Face Spaces, the system is immediately accessible and demonstrates how Gemini's intelligence combined with Veea's security creates enterprise-ready AI solutions. Business impact includes 70% analyst time savings, zero SQL errors, 100% data consistency
19 May 2026
