Ali H.

𝐋𝐨𝐨𝐤𝐢𝐧𝐠 𝐟𝐨𝐫 𝐚 Senior AI Engineer 𝐭𝐨 𝐛𝐮𝐢𝐥𝐝 𝐫𝐞𝐥𝐢𝐚𝐛𝐥𝐞 𝐚𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐚𝐠𝐞𝐧𝐭𝐢𝐜 𝐰𝐨𝐫𝐤𝐟𝐥𝐨𝐰𝐬, 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 𝐑𝐀𝐆 𝐬𝐲𝐬𝐭𝐞𝐦𝐬, 𝐚𝐧𝐝 𝐀𝐈 𝐯𝐨𝐢𝐜𝐞 𝐚𝐠𝐞𝐧𝐭𝐬? I build real-time conversational voice agents, stateful multi-agent systems with LangGraph, enterprise RAG pipelines, and custom Model Context Protocol (MCP) servers using Python. I bring 12+ years of software engineering experience building systems end to end, from architecture and development to deployment and CI/CD. Plenty of AI features work in a demo and break in production: models hallucinate numbers, leak cross-tenant data, fall prey to prompt injections hidden in files, or blow past token budgets. I focus on the engineering details that make AI dependable at scale: sub-second voice latency, deterministic guardrails, multi-tenant isolation, and token tracing. 🛠️𝐖𝐇𝐀𝐓 𝐈 𝐁𝐔𝐈𝐋𝐃 - 𝐑𝐞𝐚𝐥-𝐓𝐢𝐦𝐞 𝐀𝐈 𝐕𝐨𝐢𝐜𝐞 𝐀𝐠𝐞𝐧𝐭𝐬: Inbound customer support, outbound qualification, and appointment-booking phone bots with sub-second latency using Twilio, Vapi, Retell AI, LiveKit, and the OpenAI Realtime API. - 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 & 𝐌𝐮𝐥𝐭𝐢-𝐀𝐠𝐞𝐧𝐭 𝐒𝐲𝐬𝐭𝐞𝐦𝐬: Stateful multi-agent workflows, planner-executor loops, and reflection patterns built with LangGraph, CrewAI, and the Claude Agent SDK. - 𝐂𝐮𝐬𝐭𝐨𝐦 𝐌𝐂𝐏 𝐒𝐞𝐫𝐯𝐞𝐫𝐬 & 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧𝐬: Model Context Protocol (MCP) servers and clients that securely connect Claude, ChatGPT, Cursor, or internal agents to your databases, tools, and private APIs. - 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞 𝐑𝐀𝐆 & 𝐒𝐞𝐚𝐫𝐜𝐡: Production retrieval-augmented generation pipelines, hybrid vector + keyword search (pgvector, Pinecone, Qdrant), GraphRAG, and permission-aware retrieval over private data. - 𝐀𝐈 𝐂𝐡𝐚𝐭𝐛𝐨𝐭𝐬 & 𝐈𝐧-𝐀𝐩𝐩 𝐂𝐨𝐩𝐢𝐥𝐨𝐭𝐬: Context-aware assistants integrated into your core backend, Slack, or WhatsApp with deterministic function execution. - 𝐃𝐨𝐜𝐮𝐦𝐞𝐧𝐭 𝐈𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞 & 𝐃𝐚𝐭𝐚 𝐄𝐱𝐭𝐫𝐚𝐜𝐭𝐢𝐨𝐧: High-throughput OCR, PDF parsing, PII redaction, and schema-enforced structured data extraction using LLMs and vision models. - 𝐂𝐮𝐬𝐭𝐨𝐦 𝐌𝐋 & 𝐎𝐩𝐞𝐧-𝐒𝐨𝐮𝐫𝐜𝐞 𝐋𝐋𝐌𝐬: Feature engineering, embeddings, semantic rankers, and LoRA fine-tuning for open-source models (Llama, Mistral, Qwen) served via Ollama or vLLM. ⚡ 𝐏𝐑𝐎𝐕𝐄𝐍 𝐒𝐂𝐀𝐋𝐄 & 𝐄𝐍𝐓𝐄𝐑𝐏𝐑𝐈𝐒𝐄 𝐑𝐄𝐒𝐔𝐋𝐓𝐒 - $10𝐌 𝐑𝐚𝐢𝐬𝐞𝐝: Single-handedly built backend services, data architecture, and deployment pipelines end to end that raised $10M within months. - 𝐇𝐢𝐠𝐡-𝐏𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐀𝐏𝐈𝐬: Re-architected backend services for enterprise systems with strict latency requirements, significantly reducing API response times. - 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞 𝐀𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐨𝐧: Replaced manual operational workflows with automated real-time data pipelines, integrations, analytics, and reporting built to scale. - 𝐇𝐞𝐚𝐥𝐭𝐡𝐜𝐚𝐫𝐞 𝐏𝐥𝐚𝐭𝐟𝐨𝐫𝐦𝐬: Engineered HIPAA-conscious platforms connecting clinics and patients through secure, encrypted backend architectures and third-party integrations. 💻 𝐓𝐄𝐂𝐇 𝐒𝐓𝐀𝐂𝐊 - 𝐕𝐨𝐢𝐜𝐞 𝐀𝐈: Deepgram, ElevenLabs, Whisper, Twilio, Retell AI, Vapi, LiveKit - 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐅𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤𝐬 & 𝐌𝐂𝐏: LangGraph, LangChain, MCP (Model Context Protocol), LlamaIndex, Pydantic AI - 𝐋𝐋𝐌𝐬: OpenAI API (GPT-4o, Realtime), Anthropic Claude API, Gemini, Llama, Mistral - 𝐃𝐚𝐭𝐚𝐛𝐚𝐬𝐞𝐬 & 𝐕𝐞𝐜𝐭𝐨𝐫 𝐒𝐭𝐨𝐫𝐞𝐬: PostgreSQL (pgvector), Pinecone, Qdrant, Redis, Supabase, MongoDB - 𝐁𝐚𝐜𝐤𝐞𝐧𝐝 & 𝐈𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞: Python, FastAPI, Django, Docker, AWS, CI/CD 𝐀𝐋𝐒𝐎 𝐅𝐔𝐋𝐋-𝐒𝐓𝐀𝐂𝐊 Still full-stack: SaaS MVPs and multi-tenant platforms. I built a fintech platform on my own that went on to raise $10M, plus 5+ custom CRMs and 3+ HIPAA-conscious healthcare platforms. 💼 𝐑𝐎𝐋𝐄𝐒 𝐈 𝐅𝐈𝐋𝐋 Senior AI Engineer · Voice Agent Developer · AI Agent Architect · LangGraph / LangChain Engineer · MCP Specialist · RAG Developer · Python Backend Engineer 🤝 𝐇𝐎𝐖 𝐈 𝐖𝐎𝐑𝐊 · 𝐂𝐥𝐞𝐚𝐫, 𝐩𝐫𝐨𝐚𝐜𝐭𝐢𝐯𝐞 𝐜𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧: Regular updates so you always know the exact state of the architecture and build. · 𝐄𝐧𝐝-𝐭𝐨-𝐞𝐧𝐝 𝐨𝐰𝐧𝐞𝐫𝐬𝐡𝐢𝐩: From defining evaluation datasets to architecture, implementation, automated testing, and handover. · 𝐂𝐥𝐞𝐚𝐧, 𝐦𝐚𝐢𝐧𝐭𝐚𝐢𝐧𝐚𝐛𝐥𝐞 𝐜𝐨𝐝𝐞: Fully tested, documented, version-controlled repositories. · 𝐒𝐭𝐫𝐢𝐜𝐭 𝐜𝐨𝐧𝐟𝐢𝐝𝐞𝐧𝐭𝐢𝐚𝐥𝐢𝐭𝐲: Enterprise-grade care for data privacy, security, and IP protection. Ready to build a reliable voice agent, MCP integration, or multi-agent workflow? Click "𝐈𝐧𝐯𝐢𝐭𝐞 𝐭𝐨 𝐉𝐨𝐛" or send a message outlining your use case, data sources, and target timeline, and let's discuss the architecture, timeline, and deliverables.