Faisal I.

I build AI agents and LLM systems that hold up in production: evaluated, observable, and guarded, not just demos. 15+ years in software, 9 as an engineering manager. Today I lead 24 engineers on an IoT platform handling ~100M messages a day from 2.5M connected homes, where I built the production LLM and agentic platform from scratch. Through my consultancy I've delivered 300+ projects, including 108 on Upwork (100% Job Success, $300K+ earned), with 10+ healthcare and HIPAA clients. What I build: • AI agents & workflow automation: multi-agent orchestration with LangGraph/LangChain, MCP tool servers, human-in-the-loop approvals, Slack/Jira/CRM/API integrations. • RAG & document intelligence: hybrid retrieval (dense + sparse with rank fusion), structured extraction and grounded assistants, measured with evals so you know answer quality rather than hope for it. • Evals, guardrails & observability: benchmarking, regression evals, tracing, cost monitoring. This is where most AI projects fail, and where I spend most of my time. • LLM cost control: tiered model routing that sends each request to the cheapest model that can handle it. • AI MVPs: from scope to launch with auth, billing, dashboards and cloud deployment, architected so v2 isn't a rewrite. Selected work: • Enterprise LLM platform (IoT/energy): model orchestration with cost-aware routing, RAG on Qdrant with BGE-M3 embeddings, MCP servers, and generative workflows across Slack, Jira, Confluence and GitLab. Includes a bug-triage agent that proposes root causes. Cut engineering cycle time ~22%. • Agentic event monitoring for a fleet of millions of smart-home devices; demand-response and EV smart-charging systems live in production at 120K+ homes. • AI dental practice platform with insurance workflows (MVP in progress). • Pediatric health apps (iOS + Android) for a university health system; we are an approved vendor. • Healthcare care-services platform serving the client, their customers and their customers' customers. • AI-powered marketing engine for a real-estate SaaS platform with ~4M end customers. • Fintech data pipelines for a company later acquired by Mastercard. • Recommendation engine for Overstock. • 0→1 startup platform second engineering hire; built the core services from scratch alongside the founding CTO, then hired the team. • Regulated cloud: AWS GovCloud and Azure IL5+ deployments. Stack: Python, TypeScript/Node, LangGraph, LangChain, OpenAI, Claude, AWS Bedrock, Vertex AI, Qdrant and other vector DBs, PostgreSQL, Elasticsearch/OpenSearch, Spark, FastAPI, NextJS, AWS/GCP, Docker, Kubernetes. Best fit: you have a real workflow or dataset, need an agent or RAG system that works reliably, and want someone who will tell you what's feasible before writing code. Onshore (US) and offshore engineering capacity available. Send me the workflow you want automated and I'll reply with a feasibility take and a first-milestone plan.