class SyedAbdulKareem:
    def __init__(self):
        self.role       = "Agentic AI Developer & Team Lead"
        self.company    = "IT Automation LLC (Remote, USA)"
        self.education  = "B.E. CSE (AI & ML) — Osmania University | CGPA: 8.44"
        self.research   = "IHub-Data, IIIT-Hyderabad — EEG/BCI Neurotechnology"
        self.freelance  = "Innodata — LLM Evaluation & AI Data Annotation"
        self.building   = ["Agentic Legal AI", "Loan Origination", "Healthcare AI"]
        self.stack      = ["LangGraph", "LangChain", "RAG", "OpenAI",
                           "Anthropic", "Gemini", "LLaMA", "Groq"]
        self.focus      = ["Evals", "Observability", "Hallucination Mitigation"]
        self.passion    = "Turning cutting-edge research into reliable production AI"

I build autonomous AI systems that work in the real world — not just demos. I lead a small engineering team shipping production-grade agentic applications, and I care deeply about making agents observable, grounded, and reliable. My work spans:

• ⚖️ Legal AI — multi-agent pipelines for document analysis, case management, RAG-grounded assistance
• 💰 Financial AI — agentic loan origination, automated underwriting, risk analysis
• 🏥 Healthcare AI — medical image classification, clinical NLP, real-time inference
• 🧬 Neurotechnology — EEG-based BCI systems (OpenBCI Cyton+Daisy) at IIIT-Hyderabad, controlling devices via brain signals
• 🔍 Evals & Observability — LLM-as-Judge workflows, evaluation pipelines, trace debugging

Remote | United States · Multi-Agent Systems · RAG · LangGraph

• Lead a team of 2 engineers to design and ship autonomous agentic AI systems end-to-end using LLMs, LangGraph orchestration, RAG pipelines, and FastAPI services across three production-grade applications in the legal, financial, and public-safety domains
• Built agentic, multi-agent systems with LangGraph for autonomous workflow automation, tuned for latency, reliability, and cost
• Implemented RAG pipelines with vector databases for document understanding across structured, unstructured, and OCR-based ingestion; applied data-grounded prompting to reduce hallucination

(Specifics of these systems are company-confidential.)

Research Under Publication · Neurotechnology · Signal Processing

• Built a real-time EEG acquisition & signal-processing system using OpenBCI Cyton+Daisy with Python pipelines supporting SSVEP, Motor Imagery (MI), and P300 neural paradigms
• Optimised low-latency inference pipelines and engineered brainwave visualisation tooling with research engineers, advancing neurotechnology for autonomous human-machine interaction

Remote | Freelance

• Client: Meta SuperIntelligence Labs
• Creating high-quality datasets for state-of-the-art LLMs across text and multimodal annotation tasks
• Evaluating AI model responses for leading tech companies — providing structured feedback that directly improves model behaviour

I'm open to AI/ML full-time roles, freelancing, AI consulting, and research collaborations in legal, financial, healthcare, and neurotechnology AI domains.

"The best AI systems are the ones that make complex tasks feel simple — and the reliable ones make them trustworthy."