Open to Senior AI roles
Hey, I’m RPM 👋
RAMPRASAD MONDAL · AI + BACKEND ENGINEER · UAE
AI Engineer with a backend engineer’s discipline, 6+ years shipping systems, now building the GenAI and agentic AI platforms of tomorrow.
Work
OpenInnovation AI · AI Engineer
Dec 2025 - Jul 2026 · Abu Dhabi, UAE· 8m
Multi-agent systems · GraphRAG · air-gapped enterprise LLM deployment
- Architected a multi-tenant, multi-agent AI platform where each agent maintains an isolated Neo4j knowledge graph and GraphRAG retrieval pipeline, with dynamic query routing across agents, a knowledge-map visualisation, and a real-time active-threats dashboard.
- Built the knowledge ingestion pipeline - semantic chunking, entity extraction, and hybrid retrieval (dense vector + graph traversal) - improving response quality 50%+ and processing efficiency 50–60%, on async FastAPI + MinIO for memory-efficient document handling.
- Engineered multi-agent orchestration with structured meeting rooms for collaborative agent reasoning, plus a parallel AI report generation pipeline synthesising graph context and LLM output into stakeholder-ready deliverables.
- Delivered full English–Arabic bilingual support - a backoffice service for static translation and LLM-powered dynamic translation across agent responses, reports, and documents, with RTL throughout.
- Built a multi-modal document extraction microservice for scanned PDFs, Arabic handwritten notes, and images - high-accuracy ingestion across heterogeneous enterprise document types.
- Deployed production LLMs up to 35B parameters (including MoE) with vLLM on GPU Kubernetes, using tensor parallelism and continuous batching for enterprise-grade inference.
- Led air-gapped deployment onto NVIDIA DGX Spark, ASUS GX10, and H200 servers - zero cloud dependency, fully offline Docker installs - working directly with UAE government ministries and private-sector clients on requirements, on-site delivery, and support.
- Contributed to a TypeScript API on Hono/Bun with Zod validation and auto-generated OpenAPI docs, inside a Keycloak JWT auth + RBAC structure.
WealthyBuildwealth Technologies · Senior Software Engineer II
Jul 2020 - Nov 2025 · Abu Dhabi, UAE· 5y 5m
Fintech backend · multilingual RAG · partner & revenue platforms
- Software EngineerSoftware Engineer
- Senior SWE ISenior Software Engineer I
- Senior SWE IISenior Software Engineer II
- Built a multilingual conversational AI assistant with retrieval-augmented responses, intent classification, and entity extraction across 8+ Indian languages - partners could discover products, place orders, and get personalised recommendations.
- Developed a RAG-based Q&A service using embeddings and vector retrieval, letting users query uploaded documents and receive grounded, contextual answers.
- Led a cross-functional team of 6 delivering a production partner-payout system across engineering, product, finance, and operations.
- Scaled a Python/Django backend on AWS ECS with Redis caching and Celery/SQS processing to peak traffic of ~10K req/min, serving 250K+ registered and 20K+ daily active partners.
- Re-engineered the daily revenue pipeline - async NAV/commission/unit retrieval, then Dask + Pandas computation on a dedicated worker - cutting 2M transaction records from 10–12 hours to 3–4 minutes (99%+ faster).
- Automated invoice generation with Celery workers, Airflow orchestration, and bulk DB operations - 4 hours → 15 minutes (94% faster) for 10K+ monthly partner invoices with GST, TDS, ARN and PAN compliance.
- Built a secure authentication gateway - OAuth, OTP, Redis-backed token and session management, authorization-code exchange, request-signature validation - handling 50K+ authenticated requests daily.
- Built payment and payout infrastructure on Cashfree and Razorpay with batch transfers, partial payments, retry handling, and status reconciliation.
- Implemented centralised observability - request-ID tracing, structured logging, ClickHouse, Grafana, and automated Slack alerts - accelerating root-cause analysis across production services.
- Designed ETL pipelines for CAMS/KARVY reconciliation, improving revenue accuracy 15%.
Best Employee Award 2025Best Team Award 2025
M
MountBlue Technologies · Software Engineer Intern
Feb 2020 - Jun 2020 · Bengaluru, India· 5m
Full-stack web apps (internship)
- Built a COVID Q&A platform with user auth, comments, and real-time dashboards.
- Shipped an e-commerce platform for a local shop, end to end.
Tech Stack
Languages
PythonJavaScript
AI & agents
LLM ApplicationsAgentic AIRAG / GraphRAGDeepAgentsLangGraphLangChainKnowledge Graphs
Backend
FastAPIDjangoCeleryREST / gRPC
Data
PostgreSQLRedisNeo4jQdrantClickHouse
Cloud & infra
AWSDockerKubernetesvLLMAir-gapped Deployment
Architecture
Multi-Agent SystemsMicroservicesEvent-DrivenDistributed Systems
Recent projectsmore →
Agentic Intent Workflow
Turns a clinical note request into a typed, validated workflow of agent, deterministic and human-review tasks.
openclaude
An OpenAI-compatible API over the local Claude CLI, so agent frameworks run on a Claude login, no API key.
Second Brain OS
A local-first second brain that answers from a knowledge graph built out of your own documents, not the web.
Waste your time here 🕹️
🎮 Tic-tac-toe · you vs my minimax bot
You're X and you open. Good luck, it plays perfect minimax.
You 0 · Bot 0 · Draws 0
no games yet
⌨️ Typing test · click and type
the quick brown fox jumps over the lazy dog
WPM 0ACC 100%
The AI Lab 🧪full lab →
✏️ Draw a digit · a real neural classifier guesses it
?
logistic regression · trained on 1,797 handwritten digits · 4.7KB of weights, no server
🐍 Edit my Python · runs live in your browser
real CPython via Pyodide (WebAssembly) · loads on first run, then instant
Latest updatesprojects →
projectAgentic Intent WorkflowAug 13, 2026postDeploying Next.js to a free VM: local to live, step by stepAug 13, 2026Let’s build something
Open to Senior AI Engineer roles (also open to backend). Fastest way to reach me is email.
